{"id":32023,"name":"vendored/llama.cpp","ecosystem":"submodules","repository_url":null,"issues_count":27,"created_at":"2025-06-07T10:37:09.846Z","updated_at":"2025-06-07T10:37:09.846Z","purl":"pkg:submodules/vendored/llama.cpp","unique_repositories_count":1,"unique_repositories_count_past_30_days":1,"recent_issues":[{"uuid":"3857483195","node_id":"PR_kwDOOgWcXc6_caAF","number":48,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `8f80d1b`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2026-02-03T00:40:10.000Z","author_association":null,"state_reason":null,"created_at":"2026-01-26T20:11:06.000Z","updated_at":"2026-02-03T00:40:11.000Z","time_to_close":620944,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`8f80d1b`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `8f80d1b`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/8f80d1b254aef70a0959e314be368d05debe7294\"\u003e\u003ccode\u003e8f80d1b\u003c/code\u003e\u003c/a\u003e graph : fix nkvo offload with FA (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19105\"\u003e#19105\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/142cbe2ac68978e5dec3a2e19c1b64ef1c5740b1\"\u003e\u003ccode\u003e142cbe2\u003c/code\u003e\u003c/a\u003e ci : use new 1vCPU runner for lightweight jobs (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19107\"\u003e#19107\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/56f3ebf38ea5757447e7ea167cda15e1eebf9601\"\u003e\u003ccode\u003e56f3ebf\u003c/code\u003e\u003c/a\u003e model : add correct type for GLM 4.7 Flash (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19106\"\u003e#19106\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0c21677e43044d27f6f7a7f9f95c67f7c4b3fdb4\"\u003e\u003ccode\u003e0c21677\u003c/code\u003e\u003c/a\u003e CUDA: faster FA for GQA \u0026gt; 1 but not power of 2 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19092\"\u003e#19092\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0440bfd1605333726ea0fb7a836942660bf2f9a6\"\u003e\u003ccode\u003e0440bfd\u003c/code\u003e\u003c/a\u003e metal : fix recommendedMaxWorkingSetSize availability on legacy iOS/macOS (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/1\"\u003e#1\u003c/a\u003e...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0bf56369384ff1af2731cd1168785e283a1ba266\"\u003e\u003ccode\u003e0bf5636\u003c/code\u003e\u003c/a\u003e convert : yield Gemma3N custom_map tensors directly (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19091\"\u003e#19091\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/bcb43163aed6a8986cf3d66e90848c9c258d4936\"\u003e\u003ccode\u003ebcb4316\u003c/code\u003e\u003c/a\u003e ggml-cpu: Use tiled FA for prompt-processing (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19012\"\u003e#19012\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/d9c6ce46f747189cd6238ca7699253613f77c016\"\u003e\u003ccode\u003ed9c6ce4\u003c/code\u003e\u003c/a\u003e kv-cache : support V-less cache (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19067\"\u003e#19067\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/70d860824a72c3daaea678b74d5051fdd46d9c38\"\u003e\u003ccode\u003e70d8608\u003c/code\u003e\u003c/a\u003e convert : fix Gemma3N, GraniteMoe and Ernie4.5Moe (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19084\"\u003e#19084\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/080b161995218bb40bfc03a6446c9e2b4c7e81e0\"\u003e\u003ccode\u003e080b161\u003c/code\u003e\u003c/a\u003e completion : fix prompt cache for recurrent models (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19045\"\u003e#19045\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...8f80d1b254aef70a0959e314be368d05debe7294\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/48","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/48","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/48/packages"},{"uuid":"3682908363","node_id":"PR_kwDOOgWcXc62cpTg","number":38,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `ecf74a8`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-12-08T18:22:26.000Z","author_association":null,"state_reason":null,"created_at":"2025-12-01T21:18:39.000Z","updated_at":"2025-12-08T18:22:27.000Z","time_to_close":594227,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`ecf74a8`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `ecf74a8`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ecf74a841755fdf468d270a8727b6ed103d13344\"\u003e\u003ccode\u003eecf74a8\u003c/code\u003e\u003c/a\u003e mtmd: add mtmd_context_params::warmup option (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17652\"\u003e#17652\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/00c361fe53e5fc105a077f90a0a22d4c60936ffe\"\u003e\u003ccode\u003e00c361f\u003c/code\u003e\u003c/a\u003e fix: llama arch implementation (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17665\"\u003e#17665\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ec18edfcba94dacb166e6523612fc0129cead67a\"\u003e\u003ccode\u003eec18edf\u003c/code\u003e\u003c/a\u003e server: introduce API for serving / loading / unloading multiple models (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17470\"\u003e#17470\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/773340973473952df872b179f97b5a484a0b063d\"\u003e\u003ccode\u003e7733409\u003c/code\u003e\u003c/a\u003e common: improve verbosity level definitions (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17630\"\u003e#17630\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/cd3c1189082e9ecca172b5bea2442606f489f439\"\u003e\u003ccode\u003ecd3c118\u003c/code\u003e\u003c/a\u003e model: support Ministral3 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17644\"\u003e#17644\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/649495c9d915a284aeec5bca5d0efaa6d1bc7c87\"\u003e\u003ccode\u003e649495c\u003c/code\u003e\u003c/a\u003e metal : add FA head size 48 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17619\"\u003e#17619\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/90c72a614a6dc99064fe245b6ab85235c5f2de6a\"\u003e\u003ccode\u003e90c72a6\u003c/code\u003e\u003c/a\u003e ggml : extend the GGML_SCHED_NO_REALLOC debug logic of the scheduler (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17617\"\u003e#17617\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6eea6669125af4a6c6a8e186ba5a019152e4e868\"\u003e\u003ccode\u003e6eea666\u003c/code\u003e\u003c/a\u003e llama-graph: avoid expand_forward for fusion (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17633\"\u003e#17633\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ff90508d68de51bc4c0a1d60d636b0501cb0f3b3\"\u003e\u003ccode\u003eff90508\u003c/code\u003e\u003c/a\u003e contributing: update guidelines for AI-generated code (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17625\"\u003e#17625\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0a4aeb927db9ac28a845ceeaf1bd5e78442f0acc\"\u003e\u003ccode\u003e0a4aeb9\u003c/code\u003e\u003c/a\u003e cmake : add option to build and link LibreSSL (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17552\"\u003e#17552\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...ecf74a841755fdf468d270a8727b6ed103d13344\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/38","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/38","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/38/packages"},{"uuid":"3511094740","node_id":"PR_kwDOOgWcXc6thNsZ","number":28,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `7049736`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-10-20T18:19:44.000Z","author_association":null,"state_reason":null,"created_at":"2025-10-13T18:40:53.000Z","updated_at":"2025-10-20T18:19:45.000Z","time_to_close":603531,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`7049736`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `7049736`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/7049736b2dd9011bf819e298b844ebbc4b5afdc9\"\u003e\u003ccode\u003e7049736\u003c/code\u003e\u003c/a\u003e CUDA: fix numerical issues in tile FA kernel (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16540\"\u003e#16540\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/01d2bdc2bc61b676706830305b286b08b9885a41\"\u003e\u003ccode\u003e01d2bdc\u003c/code\u003e\u003c/a\u003e ggml : fix build broken with -march=armv9-a on MacOS (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16520\"\u003e#16520\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/56fc38b9655fbe1869d8bd6cfb269418196cea69\"\u003e\u003ccode\u003e56fc38b\u003c/code\u003e\u003c/a\u003e CANN: fix CPU memory leak in CANN backend (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16549\"\u003e#16549\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/1fb9504eb744969a990bfe4cfcf1d3d7a479541c\"\u003e\u003ccode\u003e1fb9504\u003c/code\u003e\u003c/a\u003e fix: add remark plugin to render raw HTML as literal text (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16505\"\u003e#16505\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/3f750f8d760ab5a61491e6a9409072dfeee4b4d7\"\u003e\u003ccode\u003e3f750f8\u003c/code\u003e\u003c/a\u003e metal: add support for opt_step_sgd (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16539\"\u003e#16539\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c515fc577166042234241c6bd0da9b08dcbe2bb9\"\u003e\u003ccode\u003ec515fc5\u003c/code\u003e\u003c/a\u003e ggml : fix scalar path for computing norm (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16558\"\u003e#16558\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f9bc66c3ebcfddb5f09e4b21253623caeb8e414a\"\u003e\u003ccode\u003ef9bc66c\u003c/code\u003e\u003c/a\u003e CANN: Update several operators to support FP16 data format (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16251\"\u003e#16251\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a31cf36ad946a13b3a646bf0dadf2a481e89f944\"\u003e\u003ccode\u003ea31cf36\u003c/code\u003e\u003c/a\u003e metal : add opt_step_adamw and op_sum (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16529\"\u003e#16529\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/81d54bbfd599811b354c39f04550888168be7780\"\u003e\u003ccode\u003e81d54bb\u003c/code\u003e\u003c/a\u003e webui: remove client-side context pre-check and rely on backend for limits (#...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c7be9febcbafa9af7d1b9443f86475c59c9c5f87\"\u003e\u003ccode\u003ec7be9fe\u003c/code\u003e\u003c/a\u003e [SYCL] fix UT fault cases: count-equal, argsort, pad OPs (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16521\"\u003e#16521\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...7049736b2dd9011bf819e298b844ebbc4b5afdc9\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/28","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/28","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/28/packages"},{"uuid":"2891356851","node_id":"PR_kwDOOgWcXc6sVpqz","number":27,"state":"open","title":"Bump vendored/llama.cpp from `75c91de` to `3df2244`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":0,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-10-06T18:24:41.000Z","updated_at":"2025-10-06T18:24:42.000Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`3df2244`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `3df2244`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/3df2244df40c67dfd6ad548b40ccc507a066af2b\"\u003e\u003ccode\u003e3df2244\u003c/code\u003e\u003c/a\u003e llama : add --no-host to disable host buffers (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16310\"\u003e#16310\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c08002a1988348403a5fd59b1fa3de3a10a6f92f\"\u003e\u003ccode\u003ec08002a\u003c/code\u003e\u003c/a\u003e chat : Granite Docling stopping (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16438\"\u003e#16438\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/3a002afafa8e06555f11aed88b02d055d8a166f3\"\u003e\u003ccode\u003e3a002af\u003c/code\u003e\u003c/a\u003e ci : refactor sdk caching to minimize storage (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16414\"\u003e#16414\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a23b9bdbd3b64ce172f9962249f432d01aea7437\"\u003e\u003ccode\u003ea23b9bd\u003c/code\u003e\u003c/a\u003e ggml : fix unaligned access in AMX code (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16315\"\u003e#16315\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/04e632a4aab8e6bfff0f8bc216b36ceb1e199ff9\"\u003e\u003ccode\u003e04e632a\u003c/code\u003e\u003c/a\u003e ci : remove missing reranker model files (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16444\"\u003e#16444\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a80ff183abe4e5a76316257ffa597da41b3b6fa0\"\u003e\u003ccode\u003ea80ff18\u003c/code\u003e\u003c/a\u003e ggml-cpu : fix leftover handling in ggml_vec_scale_f32 for SVE (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16443\"\u003e#16443\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/1d49ca37594fb49db6aa9518ba7c512e5ccd0108\"\u003e\u003ccode\u003e1d49ca3\u003c/code\u003e\u003c/a\u003e nix : removed metal for nix (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16118\"\u003e#16118\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c5fef0fcea3b978a2318b1af170209ecec7c37b4\"\u003e\u003ccode\u003ec5fef0f\u003c/code\u003e\u003c/a\u003e server: update readme to mention n_past_max metric (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16436\"\u003e#16436\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ca71fb9b368e3db96e028f80c4c9df6b6b370edd\"\u003e\u003ccode\u003eca71fb9\u003c/code\u003e\u003c/a\u003e model : Granite docling + Idefics3 preprocessing (SmolVLM) (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16206\"\u003e#16206\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/35266573b968e1c947b367782fb4b3eddbb4f3c0\"\u003e\u003ccode\u003e3526657\u003c/code\u003e\u003c/a\u003e ggml webgpu: actually add softmax, fix rms_norm offset (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16400\"\u003e#16400\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...3df2244df40c67dfd6ad548b40ccc507a066af2b\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/27","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/27","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/27/packages"},{"uuid":"2872136869","node_id":"PR_kwDOOgWcXc6rMVSl","number":26,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `5f7e166`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-10-06T18:24:44.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-09-29T21:36:51.000Z","updated_at":"2025-10-06T18:24:44.000Z","time_to_close":593273,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`5f7e166`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `5f7e166`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/5f7e166cbf7b9ca928c7fad990098ef32358ac75\"\u003e\u003ccode\u003e5f7e166\u003c/code\u003e\u003c/a\u003e Fix thinking blocks with quotes + add handling \u003ccode\u003e[THINK]...[/THINK]\u003c/code\u003e blocks (#...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/d72f5f7ba260b546190338b0b76f2f152581424f\"\u003e\u003ccode\u003ed72f5f7\u003c/code\u003e\u003c/a\u003e ci : add AMD runners and workflows (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16249\"\u003e#16249\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b77e6c18e1a6fac5705ed95f03af5436d67484c1\"\u003e\u003ccode\u003eb77e6c1\u003c/code\u003e\u003c/a\u003e ggml: riscv: add riscv spacemit backend (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15288\"\u003e#15288\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/2ddd3f2356c313385fafc19a9f0ce678c8fe03ee\"\u003e\u003ccode\u003e2ddd3f2\u003c/code\u003e\u003c/a\u003e sync : ggml\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/4d3d455d3c8719eb8f206de328d1b9ba191efd7c\"\u003e\u003ccode\u003e4d3d455\u003c/code\u003e\u003c/a\u003e sync : whisper.cpp (ggml/1359)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c9b1c06467aca8d30fafcab51292bcb285df5b0d\"\u003e\u003ccode\u003ec9b1c06\u003c/code\u003e\u003c/a\u003e ggml : remove -dev suffix from release version (ggml/1355)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b6ae75afb49b23db3dfbc2b01e8aabfb047e6880\"\u003e\u003ccode\u003eb6ae75a\u003c/code\u003e\u003c/a\u003e ggml : bump version to 0.9.3 (ggml/1353)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b6dff20e2fe352093039e2359370c6bb2b505e0b\"\u003e\u003ccode\u003eb6dff20\u003c/code\u003e\u003c/a\u003e ggml : prepare for development of 0.9.2-dev\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/2db78c75e4ef7497313fda6dbbb3fcaf9ca0752d\"\u003e\u003ccode\u003e2db78c7\u003c/code\u003e\u003c/a\u003e ggml : bump version to 0.9.1\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/02463ab27b1379ff5e3d936ce8b3bfd356872ea6\"\u003e\u003ccode\u003e02463ab\u003c/code\u003e\u003c/a\u003e ggml-backend : add root cause in error message if loading backend library fai...\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...5f7e166cbf7b9ca928c7fad990098ef32358ac75\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/26","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/26","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/26/packages"},{"uuid":"3442212338","node_id":"PR_kwDOOgWcXc6p6Q9a","number":25,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `351f3da`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-09-29T21:36:54.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-09-22T18:33:02.000Z","updated_at":"2025-09-29T21:36:54.000Z","time_to_close":615832,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`351f3da`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `351f3da`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/351f3da39c85f59d581fc184f09283da7f099a3b\"\u003e\u003ccode\u003e351f3da\u003c/code\u003e\u003c/a\u003e clang-tidy : disable warning about performance enum size (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16127\"\u003e#16127\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/3ecb2f671a2f49d56357f99d135a94e841759178\"\u003e\u003ccode\u003e3ecb2f6\u003c/code\u003e\u003c/a\u003e ggml : implement set_rows with i32 index (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16159\"\u003e#16159\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/432cf4304c5379bfbd1f3feed65ec205955b2f4b\"\u003e\u003ccode\u003e432cf43\u003c/code\u003e\u003c/a\u003e codeowners : update + cleanup (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16174\"\u003e#16174\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/37a23c17bdc9c99c9c6ad41168e4ced3724b72cd\"\u003e\u003ccode\u003e37a23c1\u003c/code\u003e\u003c/a\u003e common : enable \u003ccode\u003e--offline\u003c/code\u003e mode without curl support (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16137\"\u003e#16137\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/138c87ce8bd558b2cc134ada7316a3dad8eb67ac\"\u003e\u003ccode\u003e138c87c\u003c/code\u003e\u003c/a\u003e webui : fix handling incomplete chunks (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16107\"\u003e#16107\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c6db9a10278f40b4b8d3f6931728f2cce2356daa\"\u003e\u003ccode\u003ec6db9a1\u003c/code\u003e\u003c/a\u003e embedding : fix typos in README (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16171\"\u003e#16171\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/d05affbab7d1364fa7ac06c03cd33f03235ae840\"\u003e\u003ccode\u003ed05affb\u003c/code\u003e\u003c/a\u003e common : remove unused local variables (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16140\"\u003e#16140\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/4f324a556caa5be0be2261604ef4aab0faf36eb8\"\u003e\u003ccode\u003e4f324a5\u003c/code\u003e\u003c/a\u003e ggml : extend ggml_can_fuse to work with non-sequential nodes (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16123\"\u003e#16123\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a71ae3ba7a857394103f12e30b387c48242c84f2\"\u003e\u003ccode\u003ea71ae3b\u003c/code\u003e\u003c/a\u003e ggml : add ggml_op_is_empty (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16122\"\u003e#16122\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/05a2458121f9cdcc98ea6ffeeab6f23023136266\"\u003e\u003ccode\u003e05a2458\u003c/code\u003e\u003c/a\u003e codeowners : update ownership for \u003ca href=\"https://github.com/ngxson\"\u003e\u003ccode\u003e@​ngxson\u003c/code\u003e\u003c/a\u003e and \u003ca href=\"https://github.com/allozuar\"\u003e\u003ccode\u003e@​allozuar\u003c/code\u003e\u003c/a\u003e (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16128\"\u003e#16128\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...351f3da39c85f59d581fc184f09283da7f099a3b\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/25","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/25","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/25/packages"},{"uuid":"3419080786","node_id":"PR_kwDOOgWcXc6osmpF","number":24,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `b907255`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-09-22T18:33:04.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-09-15T18:41:30.000Z","updated_at":"2025-09-22T18:33:04.000Z","time_to_close":604294,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`b907255`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `b907255`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b907255f4bd169b0dc7dca9553b4c54af5170865\"\u003e\u003ccode\u003eb907255\u003c/code\u003e\u003c/a\u003e SYCL: Add COUNT_EQUAL operator support (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15991\"\u003e#15991\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/28c39da7c645185ade5436767929d7ec33006033\"\u003e\u003ccode\u003e28c39da\u003c/code\u003e\u003c/a\u003e llama-run: Fix model download on Windows (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15988\"\u003e#15988\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/106220562aca42b6738b8f51acfce0db1b8a2fb6\"\u003e\u003ccode\u003e1062205\u003c/code\u003e\u003c/a\u003e CUDA: some micro-optimizations in mmf.cuh for mul_mat_id (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15926\"\u003e#15926\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a68f31edd71cc39141113f05f7133a3e9ece8c61\"\u003e\u003ccode\u003ea68f31e\u003c/code\u003e\u003c/a\u003e fix KLD percentile output (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15999\"\u003e#15999\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b8e09f08b9a91c0401bc67d17a17c90756420346\"\u003e\u003ccode\u003eb8e09f0\u003c/code\u003e\u003c/a\u003e model : add grok-2 support (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15539\"\u003e#15539\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6c019cb04e86e2dacfe62ce7666c64e9717dde1f\"\u003e\u003ccode\u003e6c019cb\u003c/code\u003e\u003c/a\u003e server : only attempt to enable thinking if using jinja (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15967\"\u003e#15967\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/9dcd200d57bc6f05a59a6a8df361d5d183af4124\"\u003e\u003ccode\u003e9dcd200\u003c/code\u003e\u003c/a\u003e metal : remove memory pools (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15966\"\u003e#15966\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0fa154e3502e940df914f03b41475a2b80b985b0\"\u003e\u003ccode\u003e0fa154e\u003c/code\u003e\u003c/a\u003e rocm.Dockerfile: added gfx1200,gfx1201 architectures to support  AMD Radeon R...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/261e6a20ffdb79c4875e674b4f6b514bc73cff8f\"\u003e\u003ccode\u003e261e6a2\u003c/code\u003e\u003c/a\u003e Vulkan: Clean up mul_mm shader (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15987\"\u003e#15987\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a0e13dcbe5bae7025660349ef3e4ead060e507f2\"\u003e\u003ccode\u003ea0e13dc\u003c/code\u003e\u003c/a\u003e build: fix the build failures of Windows HIP release job (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15984\"\u003e#15984\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...b907255f4bd169b0dc7dca9553b4c54af5170865\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/24","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/24","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/24/packages"},{"uuid":"2809294587","node_id":"PR_kwDOOgWcXc6ncm77","number":23,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `fe1c92c`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-09-15T18:41:33.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-09-08T19:17:13.000Z","updated_at":"2025-09-15T18:41:33.000Z","time_to_close":602660,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`fe1c92c`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `fe1c92c`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/fe1c92cd7bb491e9c5767c9de413f2b7dc1e832b\"\u003e\u003ccode\u003efe1c92c\u003c/code\u003e\u003c/a\u003e media : add llama1 icon (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15878\"\u003e#15878\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/e68aa10d8f3d26fdad5b912540362d79de5460e3\"\u003e\u003ccode\u003ee68aa10\u003c/code\u003e\u003c/a\u003e vulkan: sort graph to allow more parallel execution (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15850\"\u003e#15850\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0a16bf52e6874369cb4fd2bdc4863ef984b688e7\"\u003e\u003ccode\u003e0a16bf5\u003c/code\u003e\u003c/a\u003e CUDA: generate_cu_files.py - add missing mxfp4 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15880\"\u003e#15880\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/88021565f08e0b7c4e07ac089a15ec16fae9166c\"\u003e\u003ccode\u003e8802156\u003c/code\u003e\u003c/a\u003e chat : Deepseek V3.1 reasoning and tool calling support (OpenAI Style) (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15533\"\u003e#15533\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/56920f56651908d5cce7a310dabf54ac4f6fbb7f\"\u003e\u003ccode\u003e56920f5\u003c/code\u003e\u003c/a\u003e server : bring back timings_per_token (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15879\"\u003e#15879\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b0d52998b962bd2681c34bf52af993af79f178b8\"\u003e\u003ccode\u003eb0d5299\u003c/code\u003e\u003c/a\u003e cuda : fix supports_op condition for get_rows when number of blocks is too la...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f28d4f4ac963f182ea9d0fe9e269f3f5f3782aaf\"\u003e\u003ccode\u003ef28d4f4\u003c/code\u003e\u003c/a\u003e metal : refactor + optimize (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15857\"\u003e#15857\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/9fcb29f22f5c33c04c7f0daebb24057899d67a1a\"\u003e\u003ccode\u003e9fcb29f\u003c/code\u003e\u003c/a\u003e ggml: allow casting between f32 and i32 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15783\"\u003e#15783\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/5ef22d281de9c5eaaf616874bc490b89241128cb\"\u003e\u003ccode\u003e5ef22d2\u003c/code\u003e\u003c/a\u003e CUDA: non-contiguous src0 not supported for PAD (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15869\"\u003e#15869\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/233d773d02c37982badddf3994e9953a175f34da\"\u003e\u003ccode\u003e233d773\u003c/code\u003e\u003c/a\u003e convert : force setting sliding_window from original config (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15867\"\u003e#15867\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...fe1c92cd7bb491e9c5767c9de413f2b7dc1e832b\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/23","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/23","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/23/packages"},{"uuid":"2774503454","node_id":"PR_kwDOOgWcXc6lX5Ae","number":21,"state":"open","title":"Bump vendored/llama.cpp from `75c91de` to `4d917cd`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":0,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-08-26T09:38:06.000Z","updated_at":"2025-08-26T09:38:07.000Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`4d917cd`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `4d917cd`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/4d917cd4f64cc37744e76d084659475819fb0728\"\u003e\u003ccode\u003e4d917cd\u003c/code\u003e\u003c/a\u003e vulkan: fix min subgroup 16 condition for mmid subgroup optimization (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15565\"\u003e#15565\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/886b97a5d693550c2da470c091d9d27bf38398f8\"\u003e\u003ccode\u003e886b97a\u003c/code\u003e\u003c/a\u003e tests: Generate unique input values for count_equal (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15487\"\u003e#15487\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/111f8d06f0b9169059779a35f655b343242ccbb6\"\u003e\u003ccode\u003e111f8d0\u003c/code\u003e\u003c/a\u003e metal: fix regression when no metal devices are present (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15531\"\u003e#15531\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/5eff6ec9b1220b599a43b594b1110487ab6aca08\"\u003e\u003ccode\u003e5eff6ec\u003c/code\u003e\u003c/a\u003e CUDA: MoE helper in device code, better tile sizes (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15525\"\u003e#15525\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/dfd9b5f6c7586c88588f06a644c131bec071a0a1\"\u003e\u003ccode\u003edfd9b5f\u003c/code\u003e\u003c/a\u003e model-conversion : set pooling type to none in logits.cpp (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15564\"\u003e#15564\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/5a6bc6b1a6cb665a944426c2055794950e524bf5\"\u003e\u003ccode\u003e5a6bc6b\u003c/code\u003e\u003c/a\u003e model-conversion : add model card template for embeddings [no ci] (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15557\"\u003e#15557\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6b64f74b55628e4193f4fb00313f07dbd8556528\"\u003e\u003ccode\u003e6b64f74\u003c/code\u003e\u003c/a\u003e batched-bench : fix unified KV cache handling + pp timing (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15562\"\u003e#15562\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0d5a470223fc90b6b6807921d68011ff06ae7f9e\"\u003e\u003ccode\u003e0d5a470\u003c/code\u003e\u003c/a\u003e convert : update Ernie 4.5 dense architecture name (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15555\"\u003e#15555\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b0ba31f525a2ad7fb539660be0c8f088c9869f6a\"\u003e\u003ccode\u003eb0ba31f\u003c/code\u003e\u003c/a\u003e metal : add FA kernels for HS=40 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15559\"\u003e#15559\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/7da9fed0d6f1750a8783436f6f313b87e76e6378\"\u003e\u003ccode\u003e7da9fed\u003c/code\u003e\u003c/a\u003e convert : support interns1-mini (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15412\"\u003e#15412\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...4d917cd4f64cc37744e76d084659475819fb0728\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/21","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/21","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/21/packages"},{"uuid":"2755092012","node_id":"PR_kwDOOgWcXc6kN14s","number":20,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `6d7f111`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-08-26T09:38:09.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-08-19T03:50:03.000Z","updated_at":"2025-08-26T09:38:09.000Z","time_to_close":625686,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`6d7f111`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `6d7f111`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6d7f1117e3e3285d0c5c11b5ebb0439e27920082\"\u003e\u003ccode\u003e6d7f111\u003c/code\u003e\u003c/a\u003e codeowners : remove mmv.*\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/60212f1ead2dce9bf1ac69633a7069258ae604d8\"\u003e\u003ccode\u003e60212f1\u003c/code\u003e\u003c/a\u003e sync : ggml\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f0c541d315e97b297b3421c52ebde53340ee66b3\"\u003e\u003ccode\u003ef0c541d\u003c/code\u003e\u003c/a\u003e scripts : update sync scripts\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/baa9255a45105d2d3b4ec432af13b7a6eda3ff35\"\u003e\u003ccode\u003ebaa9255\u003c/code\u003e\u003c/a\u003e llama : merge conts and reshapes and remove unnecessary cont (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15380\"\u003e#15380\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/3007baf201e7ffcda17dbdb0335997fa50a6595b\"\u003e\u003ccode\u003e3007baf\u003c/code\u003e\u003c/a\u003e readme : update hot topics (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15397\"\u003e#15397\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/d1d82416006e7ff41780cb0e9b5f28d30a267497\"\u003e\u003ccode\u003ed1d8241\u003c/code\u003e\u003c/a\u003e server : fix incoming tasks not process in order (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15395\"\u003e#15395\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/618575c5825d7d4f170e686e772178d2aae148ae\"\u003e\u003ccode\u003e618575c\u003c/code\u003e\u003c/a\u003e Fix broken build: require updated pip to support --break-system-packages (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15\"\u003e#15\u003c/a\u003e...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f44f7931729022c57319a0124931120a169e0da9\"\u003e\u003ccode\u003ef44f793\u003c/code\u003e\u003c/a\u003e ggml-quants : fix make_qp_quants NANs and IQ1 assertion errors (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15379\"\u003e#15379\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ae532eac2c1df1d8edc3d2719145895b966de1bf\"\u003e\u003ccode\u003eae532ea\u003c/code\u003e\u003c/a\u003e vulkan: disable spirv-opt for bfloat16 shaders (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15352\"\u003e#15352\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/e5155e698645242d4f019267ecc40ea9bad81b09\"\u003e\u003ccode\u003ee5155e6\u003c/code\u003e\u003c/a\u003e server : export max observed n_past value (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15361\"\u003e#15361\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...6d7f1117e3e3285d0c5c11b5ebb0439e27920082\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/20","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/20","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/20/packages"},{"uuid":"2737209725","node_id":"PR_kwDOOgWcXc6jJoF9","number":18,"state":"open","title":"Bump vendored/llama.cpp from `75c91de` to `be48528`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":0,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-08-12T00:03:47.000Z","updated_at":"2025-08-12T00:03:48.000Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`be48528`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `be48528`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/be48528b068111304e4a0bb82c028558b5705f05\"\u003e\u003ccode\u003ebe48528\u003c/code\u003e\u003c/a\u003e CANN: Add broadcast for softmax and FA (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15208\"\u003e#15208\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/cf9e5648a7ae02f7728fefcbcfd2979d83c96e8f\"\u003e\u003ccode\u003ecf9e564\u003c/code\u003e\u003c/a\u003e mtmd : Fix MinicpmV model converter and clip to avoid using hardcode. (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14750\"\u003e#14750\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/fba5c0d680b555cbac563fef57b33bc5c9621e08\"\u003e\u003ccode\u003efba5c0d\u003c/code\u003e\u003c/a\u003e chat : hotfix gpt-oss jinja raising an exception (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15243\"\u003e#15243\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/53d0a1265826f40c5dbc01d06aeab9e14fcbd69b\"\u003e\u003ccode\u003e53d0a12\u003c/code\u003e\u003c/a\u003e server : allow specifying reasoning_format in HTTP request (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15238\"\u003e#15238\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/27093afe78912494073eb043fec93a007e49653c\"\u003e\u003ccode\u003e27093af\u003c/code\u003e\u003c/a\u003e readme : update infra list (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15234\"\u003e#15234\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/228f724d9ce6c56e8cec75bfdabab4dd013def7f\"\u003e\u003ccode\u003e228f724\u003c/code\u003e\u003c/a\u003e kv-cache : fix seq_rm with seq_id == -1 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15226\"\u003e#15226\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/cd3069dfcbeee8e0e96cffb93b0ebb9e595e273a\"\u003e\u003ccode\u003ecd3069d\u003c/code\u003e\u003c/a\u003e kv-cache : log (debug) all streams in find_slot (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15176\"\u003e#15176\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/50e81bdf5db563ab57a9a722b08f96fa8a76c927\"\u003e\u003ccode\u003e50e81bd\u003c/code\u003e\u003c/a\u003e convert : fix merge conflicts (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15229\"\u003e#15229\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/1ebbaddff2a44b0599df659175d3274bd5bbeb81\"\u003e\u003ccode\u003e1ebbadd\u003c/code\u003e\u003c/a\u003e perplexity : update comments/error msg to use decode [no ci] (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15227\"\u003e#15227\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a3a7874272e5a060079658eb5cca4617b7f99062\"\u003e\u003ccode\u003ea3a7874\u003c/code\u003e\u003c/a\u003e convert : improve Mistral models integration (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14737\"\u003e#14737\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...be48528b068111304e4a0bb82c028558b5705f05\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/18","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/18","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/18/packages"},{"uuid":"2683315475","node_id":"PR_kwDOOgWcXc6f8CUT","number":16,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `6c9ee3b`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-07-21T22:42:43.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-07-21T22:22:01.000Z","updated_at":"2025-07-21T22:42:43.000Z","time_to_close":1242,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`6c9ee3b`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `6c9ee3b`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6c9ee3b17e19dcc82ab93d52ae46fdd0226d4777\"\u003e\u003ccode\u003e6c9ee3b\u003c/code\u003e\u003c/a\u003e opencl: add conv2d kernel (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14403\"\u003e#14403\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/cd465d823c378853f1f6570eebfb77f69c7e1d39\"\u003e\u003ccode\u003ecd465d8\u003c/code\u003e\u003c/a\u003e sycl: Fix im2col (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14797\"\u003e#14797\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/922042601b8a16877ccb1c2afaa2071f76734f10\"\u003e\u003ccode\u003e9220426\u003c/code\u003e\u003c/a\u003e kleidiai: add support for get_rows (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14676\"\u003e#14676\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/2ba1333b35e471b344974dde553db11bf1c2836f\"\u003e\u003ccode\u003e2ba1333\u003c/code\u003e\u003c/a\u003e docs : fix backends table in README.md (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14796\"\u003e#14796\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c2e058f1b4e799f1be085560c1bcef95b7b5ed02\"\u003e\u003ccode\u003ec2e058f\u003c/code\u003e\u003c/a\u003e vulkan/cuda: Fix im2col when KW!=KH (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14789\"\u003e#14789\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c82d48ec23fb8749c341d0838f6891fd5f6b6da0\"\u003e\u003ccode\u003ec82d48e\u003c/code\u003e\u003c/a\u003e llama : fix \u003ccode\u003e--reverse-prompt\u003c/code\u003e crashing issue (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14794\"\u003e#14794\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b4efd77f8ab407836ca73a5176f041650c5b2411\"\u003e\u003ccode\u003eb4efd77\u003c/code\u003e\u003c/a\u003e server : add parse_special option to /tokenize endpoint (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14783\"\u003e#14783\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/2be60cbc2707359241c2784f9d2e30d8fc7cdabb\"\u003e\u003ccode\u003e2be60cb\u003c/code\u003e\u003c/a\u003e docs : fix link for tools/perplexity in README.md (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14780\"\u003e#14780\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b526ad2668944a7b2b1721f60679153646313831\"\u003e\u003ccode\u003eb526ad2\u003c/code\u003e\u003c/a\u003e Documentation: Further revisions to the Vulkan section in build.md (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14785\"\u003e#14785\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/938b785764683c298e7805e712f8728489cc2f18\"\u003e\u003ccode\u003e938b785\u003c/code\u003e\u003c/a\u003e Clang-format: local files first + fix BinPacking (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14779\"\u003e#14779\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...6c9ee3b17e19dcc82ab93d52ae46fdd0226d4777\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/16","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/16","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/16/packages"},{"uuid":"2665470558","node_id":"PR_kwDOOgWcXc6e39pe","number":15,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `bdca383`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-07-21T22:22:03.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-07-14T21:36:45.000Z","updated_at":"2025-07-21T22:22:03.000Z","time_to_close":607518,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`bdca383`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `bdca383`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/bdca38376f7e8dd928defe01ce6a16218a64b040\"\u003e\u003ccode\u003ebdca383\u003c/code\u003e\u003c/a\u003e sycl: Hotfix for non dnnl codepath (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14677\"\u003e#14677\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/55c509daf51d25bfaee9c8b8ce6abff103d4473b\"\u003e\u003ccode\u003e55c509d\u003c/code\u003e\u003c/a\u003e ggml : refactor llamafile_sgemm PPC code (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14673\"\u003e#14673\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/9c9e4fc6354fc811efa06a8eb7a86d3315cec9c8\"\u003e\u003ccode\u003e9c9e4fc\u003c/code\u003e\u003c/a\u003e llama-context: add ability to get logits (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14672\"\u003e#14672\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/494c5899cb76859f32ddd913534f2685fd684a3d\"\u003e\u003ccode\u003e494c589\u003c/code\u003e\u003c/a\u003e scripts: benchmark for HTTP server throughput (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14668\"\u003e#14668\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0f4c6ec0f1a9607ba67071f8a02c69b0afc2f91e\"\u003e\u003ccode\u003e0f4c6ec\u003c/code\u003e\u003c/a\u003e SYCL: use 1D kernel for set_rows (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14618\"\u003e#14618\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/65a3ebb0aa56d6c501466e0f950ad15105fd32d8\"\u003e\u003ccode\u003e65a3ebb\u003c/code\u003e\u003c/a\u003e sycl: Batched mulmat rework for oneDNN dispatch (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14617\"\u003e#14617\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0d9226763c82562186122f3b827fa3862864a19c\"\u003e\u003ccode\u003e0d92267\u003c/code\u003e\u003c/a\u003e llama : add jinja template for rwkv-world (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14665\"\u003e#14665\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/982e347255723fe6d02e60ee30cfdd0559c884c5\"\u003e\u003ccode\u003e982e347\u003c/code\u003e\u003c/a\u003e quantize : fix minor logic flaw in --tensor-type (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14572\"\u003e#14572\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/923e3ea2e3c96a0b4c208f53bc3bc90cdcdf13c0\"\u003e\u003ccode\u003e923e3ea\u003c/code\u003e\u003c/a\u003e cuda : add set rows for bf16 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14664\"\u003e#14664\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/e743cddb60dc3a8815b9de7dd7d5c491e61b2259\"\u003e\u003ccode\u003ee743cdd\u003c/code\u003e\u003c/a\u003e cuda : add ELU support (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14657\"\u003e#14657\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...bdca38376f7e8dd928defe01ce6a16218a64b040\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/15","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/15","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/15/packages"},{"uuid":"3210489402","node_id":"PR_kwDOOgWcXc6d1qFG","number":14,"state":"open","title":"Bump vendored/llama.cpp from `0a5a3b5` to `12f55c3`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":2,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-07-07T23:13:54.000Z","updated_at":"2025-07-08T06:02:25.116Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`0a5a3b5`","new_version":"`12f55c3`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `0a5a3b5` to `12f55c3`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/12f55c302b35cfe900b84c5fe67c262026af9c44\"\u003e\u003ccode\u003e12f55c3\u003c/code\u003e\u003c/a\u003e llama : remove ggml_cont where possible (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14568\"\u003e#14568\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b9c3eefde1b67104bd993485ff38dd62abe9d70c\"\u003e\u003ccode\u003eb9c3eef\u003c/code\u003e\u003c/a\u003e CUDA: add bf16 and i32 to getrows (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14529\"\u003e#14529\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6491d6e4f1caf0ad2221865b4249ae6938a6308c\"\u003e\u003ccode\u003e6491d6e\u003c/code\u003e\u003c/a\u003e vulkan: increase LOAD_VEC_A to 8 (IQ1/IQ2) or 4 (IQ3) (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14485\"\u003e#14485\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/e592be15756ae546ba87bb5a5250b71248121971\"\u003e\u003ccode\u003ee592be1\u003c/code\u003e\u003c/a\u003e vulkan: fix rms_norm+mul fusion (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14545\"\u003e#14545\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a0374a67e2924f2e845cdc59dd67d9a44065a89c\"\u003e\u003ccode\u003ea0374a6\u003c/code\u003e\u003c/a\u003e vulkan: Handle updated FA dim2/3 definition (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14518\"\u003e#14518\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ddef99522d1ba74193b7394e803fab8db5c78bae\"\u003e\u003ccode\u003eddef995\u003c/code\u003e\u003c/a\u003e server : fix assistant prefilling when content is an array (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14360\"\u003e#14360\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/668168814601d2aef6161ce49ab186bc74177ae7\"\u003e\u003ccode\u003e6681688\u003c/code\u003e\u003c/a\u003e opencl: add GELU_ERF (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14476\"\u003e#14476\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/bac8bed248d15419137c5bc7f834582397baaebc\"\u003e\u003ccode\u003ebac8bed\u003c/code\u003e\u003c/a\u003e eval-callback : check for empty input (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14539\"\u003e#14539\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b81510a7b78f7c5a6f069c4f3d0e569b9d287913\"\u003e\u003ccode\u003eb81510a\u003c/code\u003e\u003c/a\u003e test-backend-ops: add support for specifying output format (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14368\"\u003e#14368\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ef797db357e44ecb7437fa9d22f4e1614104b342\"\u003e\u003ccode\u003eef797db\u003c/code\u003e\u003c/a\u003e metal : disable fast math in all quantize kernels (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14528\"\u003e#14528\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/0a5a3b5cdfd887cf0f8e09d9ff89dee130cfcdde...12f55c302b35cfe900b84c5fe67c262026af9c44\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/14","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/14","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/14/packages"},{"uuid":"2648085077","node_id":"PR_kwDOOgWcXc6d1pJV","number":13,"state":"open","title":"Bump vendored/llama.cpp from `0a5a3b5` to `b9c3eef`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":0,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-07-07T23:12:15.000Z","updated_at":"2025-07-07T23:12:16.000Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`0a5a3b5`","new_version":"`b9c3eef`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `0a5a3b5` to `b9c3eef`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b9c3eefde1b67104bd993485ff38dd62abe9d70c\"\u003e\u003ccode\u003eb9c3eef\u003c/code\u003e\u003c/a\u003e CUDA: add bf16 and i32 to getrows (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14529\"\u003e#14529\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6491d6e4f1caf0ad2221865b4249ae6938a6308c\"\u003e\u003ccode\u003e6491d6e\u003c/code\u003e\u003c/a\u003e vulkan: increase LOAD_VEC_A to 8 (IQ1/IQ2) or 4 (IQ3) (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14485\"\u003e#14485\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/e592be15756ae546ba87bb5a5250b71248121971\"\u003e\u003ccode\u003ee592be1\u003c/code\u003e\u003c/a\u003e vulkan: fix rms_norm+mul fusion (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14545\"\u003e#14545\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a0374a67e2924f2e845cdc59dd67d9a44065a89c\"\u003e\u003ccode\u003ea0374a6\u003c/code\u003e\u003c/a\u003e vulkan: Handle updated FA dim2/3 definition (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14518\"\u003e#14518\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ddef99522d1ba74193b7394e803fab8db5c78bae\"\u003e\u003ccode\u003eddef995\u003c/code\u003e\u003c/a\u003e server : fix assistant prefilling when content is an array (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14360\"\u003e#14360\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/668168814601d2aef6161ce49ab186bc74177ae7\"\u003e\u003ccode\u003e6681688\u003c/code\u003e\u003c/a\u003e opencl: add GELU_ERF (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14476\"\u003e#14476\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/bac8bed248d15419137c5bc7f834582397baaebc\"\u003e\u003ccode\u003ebac8bed\u003c/code\u003e\u003c/a\u003e eval-callback : check for empty input (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14539\"\u003e#14539\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b81510a7b78f7c5a6f069c4f3d0e569b9d287913\"\u003e\u003ccode\u003eb81510a\u003c/code\u003e\u003c/a\u003e test-backend-ops: add support for specifying output format (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14368\"\u003e#14368\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ef797db357e44ecb7437fa9d22f4e1614104b342\"\u003e\u003ccode\u003eef797db\u003c/code\u003e\u003c/a\u003e metal : disable fast math in all quantize kernels (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14528\"\u003e#14528\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/67d1ef23c68e1e09dc6d7423d1ef5fc0ea56ed5e\"\u003e\u003ccode\u003e67d1ef2\u003c/code\u003e\u003c/a\u003e batch : add optional for sequential equal split (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14511\"\u003e#14511\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/0a5a3b5cdfd887cf0f8e09d9ff89dee130cfcdde...b9c3eefde1b67104bd993485ff38dd62abe9d70c\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/13","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/13","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/13/packages"},{"uuid":"3190083416","node_id":"PR_kwDOOgWcXc6cx9gu","number":12,"state":"open","title":"Bump vendored/llama.cpp from `ce82bd0` to `0a5a3b5`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-06-30T22:47:09.000Z","updated_at":"2025-07-01T04:08:56.000Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`ce82bd0`","new_version":"`0a5a3b5`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"[//]: # (dependabot-start)\n⚠️  **Dependabot is rebasing this PR** ⚠️ \n\nRebasing might not happen immediately, so don't worry if this takes some time.\n\nNote: if you make any changes to this PR yourself, they will take precedence over the rebase.\n\n---\n\n[//]: # (dependabot-end)\n\nBumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `ce82bd0` to `0a5a3b5`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0a5a3b5cdfd887cf0f8e09d9ff89dee130cfcdde\"\u003e\u003ccode\u003e0a5a3b5\u003c/code\u003e\u003c/a\u003e Add Conv2d for CPU (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14388\"\u003e#14388\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/745f11fed09ba2303f3f494efb12286d382608e5\"\u003e\u003ccode\u003e745f11f\u003c/code\u003e\u003c/a\u003e memory : correctly handle failure in apply() (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14438\"\u003e#14438\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/5dd942de5922a22ec8446a4ad2203738dbcb9389\"\u003e\u003ccode\u003e5dd942d\u003c/code\u003e\u003c/a\u003e metal : disable fast-math for some cpy kernels (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14460\"\u003e#14460\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a7417f55945eb7c7a5ea6807e66564b8066a4e50\"\u003e\u003ccode\u003ea7417f5\u003c/code\u003e\u003c/a\u003e ggml-cpu: sycl: Re-enable exp f16 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14462\"\u003e#14462\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/eb3fa2913e0ace38912011fcb331624c679656f5\"\u003e\u003ccode\u003eeb3fa29\u003c/code\u003e\u003c/a\u003e test-backend-ops : disable llama test (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14461\"\u003e#14461\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c839a2da1a0d4275c3a98f70c3a7627487573a46\"\u003e\u003ccode\u003ec839a2d\u003c/code\u003e\u003c/a\u003e cmake : Remove redundant include path in CMakeLists.txt (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14452\"\u003e#14452\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/e9b6350e61d592634263a14b3d77ecbf6c1fb096\"\u003e\u003ccode\u003ee9b6350\u003c/code\u003e\u003c/a\u003e scripts : make the shell scripts cross-platform (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14341\"\u003e#14341\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/caf5681fcb47dfe9bafee94ef9aa8f669ac986c7\"\u003e\u003ccode\u003ecaf5681\u003c/code\u003e\u003c/a\u003e server : support jinja extra template kwargs (Qwen3 enable_thinking feature),...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/83790b0e7e09ab17238b16452a33053a71dbdfad\"\u003e\u003ccode\u003e83790b0\u003c/code\u003e\u003c/a\u003e server : fix appearance of the chats list context menu for Safari (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14322\"\u003e#14322\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f47c1d7106e49062279bcc57fc1077c0db61e278\"\u003e\u003ccode\u003ef47c1d7\u003c/code\u003e\u003c/a\u003e SYCL: disable faulty fp16 exp kernel (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14395\"\u003e#14395\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/ce82bd0117bd3598300b3a089d13d401b90279c7...0a5a3b5cdfd887cf0f8e09d9ff89dee130cfcdde\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/12","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/12","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/12/packages"},{"uuid":"2612741532","node_id":"PR_kwDOOgWcXc6bu0Wc","number":11,"state":"closed","title":"Bump vendored/llama.cpp from `6adc3c3` to `ce82bd0`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":3,"pull_request":true,"closed_at":"2025-06-27T15:25:16.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-06-23T20:39:49.000Z","updated_at":"2025-06-27T15:25:16.000Z","time_to_close":326727,"merged_at":"2025-06-27T15:25:16.000Z","merged_by":"dependabot[bot]","closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`6adc3c3`","new_version":"`ce82bd0`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `6adc3c3` to `ce82bd0`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ce82bd0117bd3598300b3a089d13d401b90279c7\"\u003e\u003ccode\u003ece82bd0\u003c/code\u003e\u003c/a\u003e ci: add workflow for relocatable cmake package (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14346\"\u003e#14346\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/bf2a99e3cb06a14dce2a586450d012dc31f922ae\"\u003e\u003ccode\u003ebf2a99e\u003c/code\u003e\u003c/a\u003e vulkan: update windows SDK in release.yml (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14344\"\u003e#14344\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/72c6bc3f3d0cf3bf160dddf1b803fed52bcbb0a3\"\u003e\u003ccode\u003e72c6bc3\u003c/code\u003e\u003c/a\u003e llama : better rwkv chat template and add missing \u003ccode\u003einputs.use_jinja\u003c/code\u003e setting ...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/defe2158dd5e250b4ef53994057a4ec03131a263\"\u003e\u003ccode\u003edefe215\u003c/code\u003e\u003c/a\u003e CUDA: mul_mat_v support for batch sizes \u0026gt; 1 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14262\"\u003e#14262\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/7b50d589a863c7631135c1226f6eab65cb406212\"\u003e\u003ccode\u003e7b50d58\u003c/code\u003e\u003c/a\u003e kv-cells : fix tracking of seq_pos (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14339\"\u003e#14339\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/3a9457df962b5883f2773f1c295e8c19df60d89f\"\u003e\u003ccode\u003e3a9457d\u003c/code\u003e\u003c/a\u003e vulkan: update windows SDK in CI (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14334\"\u003e#14334\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/fa4a9f2a1ccda2573189a9d4995bdf0bceb41156\"\u003e\u003ccode\u003efa4a9f2\u003c/code\u003e\u003c/a\u003e quantize : handle user-defined pruning of whole layers (blocks) (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13037\"\u003e#13037\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/238005c2dc67426cf678baa2d54c881701693288\"\u003e\u003ccode\u003e238005c\u003c/code\u003e\u003c/a\u003e gguf-py : fix SpecialVocab parsing when post_processor is null (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14330\"\u003e#14330\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/66aba7aca9a245d71f1ddf02c7a97223529752a8\"\u003e\u003ccode\u003e66aba7a\u003c/code\u003e\u003c/a\u003e run : avoid double tokenization (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14327\"\u003e#14327\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f1f5e82df6222dcbaca6396c0de44df259a1694f\"\u003e\u003ccode\u003ef1f5e82\u003c/code\u003e\u003c/a\u003e examples : fix is_first logic for tokenization (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14329\"\u003e#14329\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/6adc3c3ebc029af058ac950a8e2a825fdf18ecc6...ce82bd0117bd3598300b3a089d13d401b90279c7\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/11","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/11","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/11/packages"},{"uuid":"2579131426","node_id":"PR_kwDOOgWcXc6Zumwi","number":9,"state":"closed","title":"Bump vendored/llama.cpp from `3637576` to `7f4fbe5`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-06-16T02:43:12.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-06-09T18:31:50.000Z","updated_at":"2025-06-16T02:43:12.000Z","time_to_close":547882,"merged_at":"2025-06-16T02:43:12.000Z","merged_by":"dependabot[bot]","closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`3637576`","new_version":"`7f4fbe5`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `3637576` to `7f4fbe5`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/7f4fbe5183b23b6b2e25fd1ccc5d1fa8bb010cb7\"\u003e\u003ccode\u003e7f4fbe5\u003c/code\u003e\u003c/a\u003e llama : allow building all tests on windows when not using shared libs (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13980\"\u003e#13980\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f470bc36bed4d836b9de5a483fa0dfaee176d6f5\"\u003e\u003ccode\u003ef470bc3\u003c/code\u003e\u003c/a\u003e ggml-cpu : split arch-specific implementations (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13892\"\u003e#13892\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/8f47e25f56e9792093b7497c68e9f80bab82ed19\"\u003e\u003ccode\u003e8f47e25\u003c/code\u003e\u003c/a\u003e cuda : fix device sync on buffer clear (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14033\"\u003e#14033\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/201b31dc2e6fef3de598280b53fd3b69420a4e49\"\u003e\u003ccode\u003e201b31d\u003c/code\u003e\u003c/a\u003e graph : fix geglu (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14077\"\u003e#14077\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/e21d2d4ae26d6c5daf1602a1061aa4f8e722ca57\"\u003e\u003ccode\u003ee21d2d4\u003c/code\u003e\u003c/a\u003e CANN: Simplify the environment variable setting(\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13104\"\u003e#13104\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/dc0623fddb661926e0998538895155e4f081ff09\"\u003e\u003ccode\u003edc0623f\u003c/code\u003e\u003c/a\u003e webui: fix sidebar being covered by main content (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14082\"\u003e#14082\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/87d34b381d5868e75586210ec17b5ef5deddc276\"\u003e\u003ccode\u003e87d34b3\u003c/code\u003e\u003c/a\u003e server : fix LRU check (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14079\"\u003e#14079\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b460d16ae858c6624fd37aec316622a4060ca325\"\u003e\u003ccode\u003eb460d16\u003c/code\u003e\u003c/a\u003e sycl: Add reorder to Q6_K mmvq implementation (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13885\"\u003e#13885\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/91a8ee6a6f1f4c8547ff7b745ef95c6edc1d2af6\"\u003e\u003ccode\u003e91a8ee6\u003c/code\u003e\u003c/a\u003e add geglu activation function (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14074\"\u003e#14074\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/056eb74534ffd3efc50a9b854156eb6876a52a44\"\u003e\u003ccode\u003e056eb74\u003c/code\u003e\u003c/a\u003e CANN: Enable labeler for Ascend NPU (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13914\"\u003e#13914\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/363757628848a27a435bbf22ff9476e9aeda5f40...7f4fbe5183b23b6b2e25fd1ccc5d1fa8bb010cb7\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/9","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/9","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/9/packages"},{"uuid":"3111325223","node_id":"PR_kwDOOgWcXc6Yr_1C","number":8,"state":"open","title":"Bump vendored/llama.cpp from `6f180b9` to `3637576`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-06-02T19:14:21.000Z","updated_at":"2025-06-03T04:33:56.000Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`6f180b9`","new_version":"`3637576`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `6f180b9` to `3637576`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/363757628848a27a435bbf22ff9476e9aeda5f40\"\u003e\u003ccode\u003e3637576\u003c/code\u003e\u003c/a\u003e server : disable speculative decoding for SWA models (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13970\"\u003e#13970\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ea394d7ab1f8101716d48ce9421c94c71b93a00f\"\u003e\u003ccode\u003eea394d7\u003c/code\u003e\u003c/a\u003e metal : use F32 accumulators in FA kernels (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13975\"\u003e#13975\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/5582c49c3961269eca96822abfb87528e942dd07\"\u003e\u003ccode\u003e5582c49\u003c/code\u003e\u003c/a\u003e gemma : more consistent attention scaling for v2 and v3 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13951\"\u003e#13951\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c9bbc77931d223ed7e7cbcf1cb057bc02fd0db19\"\u003e\u003ccode\u003ec9bbc77\u003c/code\u003e\u003c/a\u003e \u003ccode\u003eserver\u003c/code\u003e: update deepseek reasoning format (pass reasoning_content as diffs) ...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/bfd322796cd838f906535ff3352624fc46338894\"\u003e\u003ccode\u003ebfd3227\u003c/code\u003e\u003c/a\u003e mtmd : fix memory leak in mtmd_helper_eval_chunk_single (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13961\"\u003e#13961\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/093e3f1feb16e25e58f7d61e01266c830dd424b8\"\u003e\u003ccode\u003e093e3f1\u003c/code\u003e\u003c/a\u003e cmake : Handle mixed-case 'Power' strings in POWER CPU detection (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13966\"\u003e#13966\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/663445b0deb21fb602176da030d4154197a4fca6\"\u003e\u003ccode\u003e663445b\u003c/code\u003e\u003c/a\u003e sycl: quantize and reorder the input to q8_1 when reorder is enabled (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13826\"\u003e#13826\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/7675c555a13c9f473249e59a54db35032ce8e0fc\"\u003e\u003ccode\u003e7675c55\u003c/code\u003e\u003c/a\u003e gguf: fix failure on version == 0 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13956\"\u003e#13956\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/5e1c3aed4074480f63e914d6c44c93536ed1452a\"\u003e\u003ccode\u003e5e1c3ae\u003c/code\u003e\u003c/a\u003e convert : fix nomic-bert-moe mask token (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13757\"\u003e#13757\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c496fe0b1da28618dd17bda8b0a6bf5004554080\"\u003e\u003ccode\u003ec496fe0\u003c/code\u003e\u003c/a\u003e convert : fix vocab padding code for bert models (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13954\"\u003e#13954\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/6f180b915c9ed9ec0c240b5dcd64644988fb5e82...363757628848a27a435bbf22ff9476e9aeda5f40\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/8","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/8","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/8/packages"},{"uuid":"3091869130","node_id":"PR_kwDOOgWcXc6XrJEs","number":7,"state":"open","title":"Bump vendored/llama.cpp from `6a2bc8b` to `6f180b9`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-05-26T18:48:01.000Z","updated_at":"2025-05-28T03:06:19.000Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`6a2bc8b`","new_version":"`6f180b9`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `6a2bc8b` to `6f180b9`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6f180b915c9ed9ec0c240b5dcd64644988fb5e82\"\u003e\u003ccode\u003e6f180b9\u003c/code\u003e\u003c/a\u003e SYCL: Add non contiguous support in RMS_NORM and NORM kernels (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13611\"\u003e#13611\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/03f582ae8fccecff225c30a2802461b44761e822\"\u003e\u003ccode\u003e03f582a\u003c/code\u003e\u003c/a\u003e server: fix streaming crashes (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13786\"\u003e#13786\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/88c125f2acce0e25e5fc8481ab0681415fc64a10\"\u003e\u003ccode\u003e88c125f\u003c/code\u003e\u003c/a\u003e examples/training: Fix file name in README (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13803\"\u003e#13803\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/d74e94c1b3ac02db01e47008e1f8d371029d81ac\"\u003e\u003ccode\u003ed74e94c\u003c/code\u003e\u003c/a\u003e \u003ccode\u003eserver\u003c/code\u003e: fix format of streamed tool call deltas (diff name, fix id location...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f13847cfb560d92492b480cca2c5d6aa9473cde3\"\u003e\u003ccode\u003ef13847c\u003c/code\u003e\u003c/a\u003e server: fix regression on streamed non-chat completion w/ stops (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13785\"\u003e#13785\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/79c137f77677b3c8ee3c60a7da033721b938399a\"\u003e\u003ccode\u003e79c137f\u003c/code\u003e\u003c/a\u003e examples : allow extracting embeddings from decoder contexts (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13797\"\u003e#13797\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/22229314fc46b2f741bb21b12cde71f6c6a60b52\"\u003e\u003ccode\u003e2222931\u003c/code\u003e\u003c/a\u003e llama : clarify deprecation message (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13794\"\u003e#13794\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/9012eb9b454a82eaa4cd77ae904c0ea391e4db42\"\u003e\u003ccode\u003e9012eb9\u003c/code\u003e\u003c/a\u003e sycl: Add more debug prints (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13640\"\u003e#13640\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/fef693dc6b959a8e8ba11558fbeaad0b264dd457\"\u003e\u003ccode\u003efef693d\u003c/code\u003e\u003c/a\u003e vulkan: mark IM2COL as supporting non-contig (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13783\"\u003e#13783\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/2d38b6e4004fb1c341723e657fb1e71a4d3fb473\"\u003e\u003ccode\u003e2d38b6e\u003c/code\u003e\u003c/a\u003e CANN: Add the basic supports of Flash Attention kernel (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13627\"\u003e#13627\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/6a2bc8bfb7cd502e5ebc72e36c97a6f848c21c2c...6f180b915c9ed9ec0c240b5dcd64644988fb5e82\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/7","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/7","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/7/packages"}],"issue_packages":[{"old_version":"`75c91de`","new_version":"`8f80d1b`","update_type":null,"path":null,"pr_created_at":"2026-01-26T20:11:06.000Z","version_change":"`75c91de` → `8f80d1b`","issue":{"uuid":"3857483195","node_id":"PR_kwDOOgWcXc6_caAF","number":48,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `8f80d1b`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2026-02-03T00:40:10.000Z","author_association":null,"state_reason":null,"created_at":"2026-01-26T20:11:06.000Z","updated_at":"2026-02-03T00:40:11.000Z","time_to_close":620944,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`8f80d1b`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `8f80d1b`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/8f80d1b254aef70a0959e314be368d05debe7294\"\u003e\u003ccode\u003e8f80d1b\u003c/code\u003e\u003c/a\u003e graph : fix nkvo offload with FA (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19105\"\u003e#19105\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/142cbe2ac68978e5dec3a2e19c1b64ef1c5740b1\"\u003e\u003ccode\u003e142cbe2\u003c/code\u003e\u003c/a\u003e ci : use new 1vCPU runner for lightweight jobs (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19107\"\u003e#19107\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/56f3ebf38ea5757447e7ea167cda15e1eebf9601\"\u003e\u003ccode\u003e56f3ebf\u003c/code\u003e\u003c/a\u003e model : add correct type for GLM 4.7 Flash (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19106\"\u003e#19106\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0c21677e43044d27f6f7a7f9f95c67f7c4b3fdb4\"\u003e\u003ccode\u003e0c21677\u003c/code\u003e\u003c/a\u003e CUDA: faster FA for GQA \u0026gt; 1 but not power of 2 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19092\"\u003e#19092\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0440bfd1605333726ea0fb7a836942660bf2f9a6\"\u003e\u003ccode\u003e0440bfd\u003c/code\u003e\u003c/a\u003e metal : fix recommendedMaxWorkingSetSize availability on legacy iOS/macOS (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/1\"\u003e#1\u003c/a\u003e...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0bf56369384ff1af2731cd1168785e283a1ba266\"\u003e\u003ccode\u003e0bf5636\u003c/code\u003e\u003c/a\u003e convert : yield Gemma3N custom_map tensors directly (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19091\"\u003e#19091\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/bcb43163aed6a8986cf3d66e90848c9c258d4936\"\u003e\u003ccode\u003ebcb4316\u003c/code\u003e\u003c/a\u003e ggml-cpu: Use tiled FA for prompt-processing (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19012\"\u003e#19012\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/d9c6ce46f747189cd6238ca7699253613f77c016\"\u003e\u003ccode\u003ed9c6ce4\u003c/code\u003e\u003c/a\u003e kv-cache : support V-less cache (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19067\"\u003e#19067\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/70d860824a72c3daaea678b74d5051fdd46d9c38\"\u003e\u003ccode\u003e70d8608\u003c/code\u003e\u003c/a\u003e convert : fix Gemma3N, GraniteMoe and Ernie4.5Moe (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19084\"\u003e#19084\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/080b161995218bb40bfc03a6446c9e2b4c7e81e0\"\u003e\u003ccode\u003e080b161\u003c/code\u003e\u003c/a\u003e completion : fix prompt cache for recurrent models (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/19045\"\u003e#19045\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...8f80d1b254aef70a0959e314be368d05debe7294\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/48","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/48","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/48/packages"}},{"old_version":"`75c91de`","new_version":"`ecf74a8`","update_type":null,"path":null,"pr_created_at":"2025-12-01T21:18:39.000Z","version_change":"`75c91de` → `ecf74a8`","issue":{"uuid":"3682908363","node_id":"PR_kwDOOgWcXc62cpTg","number":38,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `ecf74a8`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-12-08T18:22:26.000Z","author_association":null,"state_reason":null,"created_at":"2025-12-01T21:18:39.000Z","updated_at":"2025-12-08T18:22:27.000Z","time_to_close":594227,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`ecf74a8`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `ecf74a8`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ecf74a841755fdf468d270a8727b6ed103d13344\"\u003e\u003ccode\u003eecf74a8\u003c/code\u003e\u003c/a\u003e mtmd: add mtmd_context_params::warmup option (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17652\"\u003e#17652\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/00c361fe53e5fc105a077f90a0a22d4c60936ffe\"\u003e\u003ccode\u003e00c361f\u003c/code\u003e\u003c/a\u003e fix: llama arch implementation (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17665\"\u003e#17665\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ec18edfcba94dacb166e6523612fc0129cead67a\"\u003e\u003ccode\u003eec18edf\u003c/code\u003e\u003c/a\u003e server: introduce API for serving / loading / unloading multiple models (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17470\"\u003e#17470\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/773340973473952df872b179f97b5a484a0b063d\"\u003e\u003ccode\u003e7733409\u003c/code\u003e\u003c/a\u003e common: improve verbosity level definitions (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17630\"\u003e#17630\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/cd3c1189082e9ecca172b5bea2442606f489f439\"\u003e\u003ccode\u003ecd3c118\u003c/code\u003e\u003c/a\u003e model: support Ministral3 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17644\"\u003e#17644\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/649495c9d915a284aeec5bca5d0efaa6d1bc7c87\"\u003e\u003ccode\u003e649495c\u003c/code\u003e\u003c/a\u003e metal : add FA head size 48 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17619\"\u003e#17619\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/90c72a614a6dc99064fe245b6ab85235c5f2de6a\"\u003e\u003ccode\u003e90c72a6\u003c/code\u003e\u003c/a\u003e ggml : extend the GGML_SCHED_NO_REALLOC debug logic of the scheduler (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17617\"\u003e#17617\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6eea6669125af4a6c6a8e186ba5a019152e4e868\"\u003e\u003ccode\u003e6eea666\u003c/code\u003e\u003c/a\u003e llama-graph: avoid expand_forward for fusion (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17633\"\u003e#17633\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ff90508d68de51bc4c0a1d60d636b0501cb0f3b3\"\u003e\u003ccode\u003eff90508\u003c/code\u003e\u003c/a\u003e contributing: update guidelines for AI-generated code (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17625\"\u003e#17625\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0a4aeb927db9ac28a845ceeaf1bd5e78442f0acc\"\u003e\u003ccode\u003e0a4aeb9\u003c/code\u003e\u003c/a\u003e cmake : add option to build and link LibreSSL (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/17552\"\u003e#17552\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...ecf74a841755fdf468d270a8727b6ed103d13344\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/38","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/38","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/38/packages"}},{"old_version":"`75c91de`","new_version":"`7049736`","update_type":null,"path":null,"pr_created_at":"2025-10-13T18:40:53.000Z","version_change":"`75c91de` → `7049736`","issue":{"uuid":"3511094740","node_id":"PR_kwDOOgWcXc6thNsZ","number":28,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `7049736`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-10-20T18:19:44.000Z","author_association":null,"state_reason":null,"created_at":"2025-10-13T18:40:53.000Z","updated_at":"2025-10-20T18:19:45.000Z","time_to_close":603531,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`7049736`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `7049736`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/7049736b2dd9011bf819e298b844ebbc4b5afdc9\"\u003e\u003ccode\u003e7049736\u003c/code\u003e\u003c/a\u003e CUDA: fix numerical issues in tile FA kernel (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16540\"\u003e#16540\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/01d2bdc2bc61b676706830305b286b08b9885a41\"\u003e\u003ccode\u003e01d2bdc\u003c/code\u003e\u003c/a\u003e ggml : fix build broken with -march=armv9-a on MacOS (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16520\"\u003e#16520\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/56fc38b9655fbe1869d8bd6cfb269418196cea69\"\u003e\u003ccode\u003e56fc38b\u003c/code\u003e\u003c/a\u003e CANN: fix CPU memory leak in CANN backend (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16549\"\u003e#16549\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/1fb9504eb744969a990bfe4cfcf1d3d7a479541c\"\u003e\u003ccode\u003e1fb9504\u003c/code\u003e\u003c/a\u003e fix: add remark plugin to render raw HTML as literal text (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16505\"\u003e#16505\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/3f750f8d760ab5a61491e6a9409072dfeee4b4d7\"\u003e\u003ccode\u003e3f750f8\u003c/code\u003e\u003c/a\u003e metal: add support for opt_step_sgd (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16539\"\u003e#16539\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c515fc577166042234241c6bd0da9b08dcbe2bb9\"\u003e\u003ccode\u003ec515fc5\u003c/code\u003e\u003c/a\u003e ggml : fix scalar path for computing norm (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16558\"\u003e#16558\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f9bc66c3ebcfddb5f09e4b21253623caeb8e414a\"\u003e\u003ccode\u003ef9bc66c\u003c/code\u003e\u003c/a\u003e CANN: Update several operators to support FP16 data format (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16251\"\u003e#16251\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a31cf36ad946a13b3a646bf0dadf2a481e89f944\"\u003e\u003ccode\u003ea31cf36\u003c/code\u003e\u003c/a\u003e metal : add opt_step_adamw and op_sum (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16529\"\u003e#16529\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/81d54bbfd599811b354c39f04550888168be7780\"\u003e\u003ccode\u003e81d54bb\u003c/code\u003e\u003c/a\u003e webui: remove client-side context pre-check and rely on backend for limits (#...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c7be9febcbafa9af7d1b9443f86475c59c9c5f87\"\u003e\u003ccode\u003ec7be9fe\u003c/code\u003e\u003c/a\u003e [SYCL] fix UT fault cases: count-equal, argsort, pad OPs (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16521\"\u003e#16521\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...7049736b2dd9011bf819e298b844ebbc4b5afdc9\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/28","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/28","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/28/packages"}},{"old_version":"`75c91de`","new_version":"`3df2244`","update_type":null,"path":null,"pr_created_at":"2025-10-06T18:24:41.000Z","version_change":"`75c91de` → `3df2244`","issue":{"uuid":"2891356851","node_id":"PR_kwDOOgWcXc6sVpqz","number":27,"state":"open","title":"Bump vendored/llama.cpp from `75c91de` to `3df2244`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":0,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-10-06T18:24:41.000Z","updated_at":"2025-10-06T18:24:42.000Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`3df2244`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `3df2244`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/3df2244df40c67dfd6ad548b40ccc507a066af2b\"\u003e\u003ccode\u003e3df2244\u003c/code\u003e\u003c/a\u003e llama : add --no-host to disable host buffers (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16310\"\u003e#16310\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c08002a1988348403a5fd59b1fa3de3a10a6f92f\"\u003e\u003ccode\u003ec08002a\u003c/code\u003e\u003c/a\u003e chat : Granite Docling stopping (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16438\"\u003e#16438\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/3a002afafa8e06555f11aed88b02d055d8a166f3\"\u003e\u003ccode\u003e3a002af\u003c/code\u003e\u003c/a\u003e ci : refactor sdk caching to minimize storage (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16414\"\u003e#16414\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a23b9bdbd3b64ce172f9962249f432d01aea7437\"\u003e\u003ccode\u003ea23b9bd\u003c/code\u003e\u003c/a\u003e ggml : fix unaligned access in AMX code (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16315\"\u003e#16315\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/04e632a4aab8e6bfff0f8bc216b36ceb1e199ff9\"\u003e\u003ccode\u003e04e632a\u003c/code\u003e\u003c/a\u003e ci : remove missing reranker model files (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16444\"\u003e#16444\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a80ff183abe4e5a76316257ffa597da41b3b6fa0\"\u003e\u003ccode\u003ea80ff18\u003c/code\u003e\u003c/a\u003e ggml-cpu : fix leftover handling in ggml_vec_scale_f32 for SVE (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16443\"\u003e#16443\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/1d49ca37594fb49db6aa9518ba7c512e5ccd0108\"\u003e\u003ccode\u003e1d49ca3\u003c/code\u003e\u003c/a\u003e nix : removed metal for nix (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16118\"\u003e#16118\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c5fef0fcea3b978a2318b1af170209ecec7c37b4\"\u003e\u003ccode\u003ec5fef0f\u003c/code\u003e\u003c/a\u003e server: update readme to mention n_past_max metric (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16436\"\u003e#16436\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ca71fb9b368e3db96e028f80c4c9df6b6b370edd\"\u003e\u003ccode\u003eca71fb9\u003c/code\u003e\u003c/a\u003e model : Granite docling + Idefics3 preprocessing (SmolVLM) (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16206\"\u003e#16206\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/35266573b968e1c947b367782fb4b3eddbb4f3c0\"\u003e\u003ccode\u003e3526657\u003c/code\u003e\u003c/a\u003e ggml webgpu: actually add softmax, fix rms_norm offset (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16400\"\u003e#16400\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...3df2244df40c67dfd6ad548b40ccc507a066af2b\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/27","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/27","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/27/packages"}},{"old_version":"`75c91de`","new_version":"`5f7e166`","update_type":null,"path":null,"pr_created_at":"2025-09-29T21:36:51.000Z","version_change":"`75c91de` → `5f7e166`","issue":{"uuid":"2872136869","node_id":"PR_kwDOOgWcXc6rMVSl","number":26,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `5f7e166`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-10-06T18:24:44.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-09-29T21:36:51.000Z","updated_at":"2025-10-06T18:24:44.000Z","time_to_close":593273,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`5f7e166`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `5f7e166`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/5f7e166cbf7b9ca928c7fad990098ef32358ac75\"\u003e\u003ccode\u003e5f7e166\u003c/code\u003e\u003c/a\u003e Fix thinking blocks with quotes + add handling \u003ccode\u003e[THINK]...[/THINK]\u003c/code\u003e blocks (#...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/d72f5f7ba260b546190338b0b76f2f152581424f\"\u003e\u003ccode\u003ed72f5f7\u003c/code\u003e\u003c/a\u003e ci : add AMD runners and workflows (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16249\"\u003e#16249\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b77e6c18e1a6fac5705ed95f03af5436d67484c1\"\u003e\u003ccode\u003eb77e6c1\u003c/code\u003e\u003c/a\u003e ggml: riscv: add riscv spacemit backend (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15288\"\u003e#15288\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/2ddd3f2356c313385fafc19a9f0ce678c8fe03ee\"\u003e\u003ccode\u003e2ddd3f2\u003c/code\u003e\u003c/a\u003e sync : ggml\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/4d3d455d3c8719eb8f206de328d1b9ba191efd7c\"\u003e\u003ccode\u003e4d3d455\u003c/code\u003e\u003c/a\u003e sync : whisper.cpp (ggml/1359)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c9b1c06467aca8d30fafcab51292bcb285df5b0d\"\u003e\u003ccode\u003ec9b1c06\u003c/code\u003e\u003c/a\u003e ggml : remove -dev suffix from release version (ggml/1355)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b6ae75afb49b23db3dfbc2b01e8aabfb047e6880\"\u003e\u003ccode\u003eb6ae75a\u003c/code\u003e\u003c/a\u003e ggml : bump version to 0.9.3 (ggml/1353)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b6dff20e2fe352093039e2359370c6bb2b505e0b\"\u003e\u003ccode\u003eb6dff20\u003c/code\u003e\u003c/a\u003e ggml : prepare for development of 0.9.2-dev\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/2db78c75e4ef7497313fda6dbbb3fcaf9ca0752d\"\u003e\u003ccode\u003e2db78c7\u003c/code\u003e\u003c/a\u003e ggml : bump version to 0.9.1\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/02463ab27b1379ff5e3d936ce8b3bfd356872ea6\"\u003e\u003ccode\u003e02463ab\u003c/code\u003e\u003c/a\u003e ggml-backend : add root cause in error message if loading backend library fai...\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...5f7e166cbf7b9ca928c7fad990098ef32358ac75\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/26","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/26","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/26/packages"}},{"old_version":"`75c91de`","new_version":"`351f3da`","update_type":null,"path":null,"pr_created_at":"2025-09-22T18:33:02.000Z","version_change":"`75c91de` → `351f3da`","issue":{"uuid":"3442212338","node_id":"PR_kwDOOgWcXc6p6Q9a","number":25,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `351f3da`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-09-29T21:36:54.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-09-22T18:33:02.000Z","updated_at":"2025-09-29T21:36:54.000Z","time_to_close":615832,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`351f3da`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `351f3da`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/351f3da39c85f59d581fc184f09283da7f099a3b\"\u003e\u003ccode\u003e351f3da\u003c/code\u003e\u003c/a\u003e clang-tidy : disable warning about performance enum size (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16127\"\u003e#16127\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/3ecb2f671a2f49d56357f99d135a94e841759178\"\u003e\u003ccode\u003e3ecb2f6\u003c/code\u003e\u003c/a\u003e ggml : implement set_rows with i32 index (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16159\"\u003e#16159\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/432cf4304c5379bfbd1f3feed65ec205955b2f4b\"\u003e\u003ccode\u003e432cf43\u003c/code\u003e\u003c/a\u003e codeowners : update + cleanup (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16174\"\u003e#16174\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/37a23c17bdc9c99c9c6ad41168e4ced3724b72cd\"\u003e\u003ccode\u003e37a23c1\u003c/code\u003e\u003c/a\u003e common : enable \u003ccode\u003e--offline\u003c/code\u003e mode without curl support (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16137\"\u003e#16137\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/138c87ce8bd558b2cc134ada7316a3dad8eb67ac\"\u003e\u003ccode\u003e138c87c\u003c/code\u003e\u003c/a\u003e webui : fix handling incomplete chunks (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16107\"\u003e#16107\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c6db9a10278f40b4b8d3f6931728f2cce2356daa\"\u003e\u003ccode\u003ec6db9a1\u003c/code\u003e\u003c/a\u003e embedding : fix typos in README (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16171\"\u003e#16171\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/d05affbab7d1364fa7ac06c03cd33f03235ae840\"\u003e\u003ccode\u003ed05affb\u003c/code\u003e\u003c/a\u003e common : remove unused local variables (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16140\"\u003e#16140\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/4f324a556caa5be0be2261604ef4aab0faf36eb8\"\u003e\u003ccode\u003e4f324a5\u003c/code\u003e\u003c/a\u003e ggml : extend ggml_can_fuse to work with non-sequential nodes (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16123\"\u003e#16123\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a71ae3ba7a857394103f12e30b387c48242c84f2\"\u003e\u003ccode\u003ea71ae3b\u003c/code\u003e\u003c/a\u003e ggml : add ggml_op_is_empty (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16122\"\u003e#16122\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/05a2458121f9cdcc98ea6ffeeab6f23023136266\"\u003e\u003ccode\u003e05a2458\u003c/code\u003e\u003c/a\u003e codeowners : update ownership for \u003ca href=\"https://github.com/ngxson\"\u003e\u003ccode\u003e@​ngxson\u003c/code\u003e\u003c/a\u003e and \u003ca href=\"https://github.com/allozuar\"\u003e\u003ccode\u003e@​allozuar\u003c/code\u003e\u003c/a\u003e (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/16128\"\u003e#16128\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...351f3da39c85f59d581fc184f09283da7f099a3b\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/25","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/25","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/25/packages"}},{"old_version":"`75c91de`","new_version":"`b907255`","update_type":null,"path":null,"pr_created_at":"2025-09-15T18:41:30.000Z","version_change":"`75c91de` → `b907255`","issue":{"uuid":"3419080786","node_id":"PR_kwDOOgWcXc6osmpF","number":24,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `b907255`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-09-22T18:33:04.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-09-15T18:41:30.000Z","updated_at":"2025-09-22T18:33:04.000Z","time_to_close":604294,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`b907255`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `b907255`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b907255f4bd169b0dc7dca9553b4c54af5170865\"\u003e\u003ccode\u003eb907255\u003c/code\u003e\u003c/a\u003e SYCL: Add COUNT_EQUAL operator support (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15991\"\u003e#15991\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/28c39da7c645185ade5436767929d7ec33006033\"\u003e\u003ccode\u003e28c39da\u003c/code\u003e\u003c/a\u003e llama-run: Fix model download on Windows (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15988\"\u003e#15988\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/106220562aca42b6738b8f51acfce0db1b8a2fb6\"\u003e\u003ccode\u003e1062205\u003c/code\u003e\u003c/a\u003e CUDA: some micro-optimizations in mmf.cuh for mul_mat_id (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15926\"\u003e#15926\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a68f31edd71cc39141113f05f7133a3e9ece8c61\"\u003e\u003ccode\u003ea68f31e\u003c/code\u003e\u003c/a\u003e fix KLD percentile output (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15999\"\u003e#15999\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b8e09f08b9a91c0401bc67d17a17c90756420346\"\u003e\u003ccode\u003eb8e09f0\u003c/code\u003e\u003c/a\u003e model : add grok-2 support (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15539\"\u003e#15539\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6c019cb04e86e2dacfe62ce7666c64e9717dde1f\"\u003e\u003ccode\u003e6c019cb\u003c/code\u003e\u003c/a\u003e server : only attempt to enable thinking if using jinja (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15967\"\u003e#15967\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/9dcd200d57bc6f05a59a6a8df361d5d183af4124\"\u003e\u003ccode\u003e9dcd200\u003c/code\u003e\u003c/a\u003e metal : remove memory pools (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15966\"\u003e#15966\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0fa154e3502e940df914f03b41475a2b80b985b0\"\u003e\u003ccode\u003e0fa154e\u003c/code\u003e\u003c/a\u003e rocm.Dockerfile: added gfx1200,gfx1201 architectures to support  AMD Radeon R...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/261e6a20ffdb79c4875e674b4f6b514bc73cff8f\"\u003e\u003ccode\u003e261e6a2\u003c/code\u003e\u003c/a\u003e Vulkan: Clean up mul_mm shader (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15987\"\u003e#15987\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a0e13dcbe5bae7025660349ef3e4ead060e507f2\"\u003e\u003ccode\u003ea0e13dc\u003c/code\u003e\u003c/a\u003e build: fix the build failures of Windows HIP release job (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15984\"\u003e#15984\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...b907255f4bd169b0dc7dca9553b4c54af5170865\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/24","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/24","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/24/packages"}},{"old_version":"`75c91de`","new_version":"`fe1c92c`","update_type":null,"path":null,"pr_created_at":"2025-09-08T19:17:13.000Z","version_change":"`75c91de` → `fe1c92c`","issue":{"uuid":"2809294587","node_id":"PR_kwDOOgWcXc6ncm77","number":23,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `fe1c92c`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-09-15T18:41:33.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-09-08T19:17:13.000Z","updated_at":"2025-09-15T18:41:33.000Z","time_to_close":602660,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`fe1c92c`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `fe1c92c`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/fe1c92cd7bb491e9c5767c9de413f2b7dc1e832b\"\u003e\u003ccode\u003efe1c92c\u003c/code\u003e\u003c/a\u003e media : add llama1 icon (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15878\"\u003e#15878\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/e68aa10d8f3d26fdad5b912540362d79de5460e3\"\u003e\u003ccode\u003ee68aa10\u003c/code\u003e\u003c/a\u003e vulkan: sort graph to allow more parallel execution (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15850\"\u003e#15850\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0a16bf52e6874369cb4fd2bdc4863ef984b688e7\"\u003e\u003ccode\u003e0a16bf5\u003c/code\u003e\u003c/a\u003e CUDA: generate_cu_files.py - add missing mxfp4 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15880\"\u003e#15880\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/88021565f08e0b7c4e07ac089a15ec16fae9166c\"\u003e\u003ccode\u003e8802156\u003c/code\u003e\u003c/a\u003e chat : Deepseek V3.1 reasoning and tool calling support (OpenAI Style) (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15533\"\u003e#15533\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/56920f56651908d5cce7a310dabf54ac4f6fbb7f\"\u003e\u003ccode\u003e56920f5\u003c/code\u003e\u003c/a\u003e server : bring back timings_per_token (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15879\"\u003e#15879\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b0d52998b962bd2681c34bf52af993af79f178b8\"\u003e\u003ccode\u003eb0d5299\u003c/code\u003e\u003c/a\u003e cuda : fix supports_op condition for get_rows when number of blocks is too la...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f28d4f4ac963f182ea9d0fe9e269f3f5f3782aaf\"\u003e\u003ccode\u003ef28d4f4\u003c/code\u003e\u003c/a\u003e metal : refactor + optimize (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15857\"\u003e#15857\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/9fcb29f22f5c33c04c7f0daebb24057899d67a1a\"\u003e\u003ccode\u003e9fcb29f\u003c/code\u003e\u003c/a\u003e ggml: allow casting between f32 and i32 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15783\"\u003e#15783\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/5ef22d281de9c5eaaf616874bc490b89241128cb\"\u003e\u003ccode\u003e5ef22d2\u003c/code\u003e\u003c/a\u003e CUDA: non-contiguous src0 not supported for PAD (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15869\"\u003e#15869\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/233d773d02c37982badddf3994e9953a175f34da\"\u003e\u003ccode\u003e233d773\u003c/code\u003e\u003c/a\u003e convert : force setting sliding_window from original config (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15867\"\u003e#15867\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...fe1c92cd7bb491e9c5767c9de413f2b7dc1e832b\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/23","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/23","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/23/packages"}},{"old_version":"`75c91de`","new_version":"`4d917cd`","update_type":null,"path":null,"pr_created_at":"2025-08-26T09:38:06.000Z","version_change":"`75c91de` → `4d917cd`","issue":{"uuid":"2774503454","node_id":"PR_kwDOOgWcXc6lX5Ae","number":21,"state":"open","title":"Bump vendored/llama.cpp from `75c91de` to `4d917cd`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":0,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-08-26T09:38:06.000Z","updated_at":"2025-08-26T09:38:07.000Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`4d917cd`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `4d917cd`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/4d917cd4f64cc37744e76d084659475819fb0728\"\u003e\u003ccode\u003e4d917cd\u003c/code\u003e\u003c/a\u003e vulkan: fix min subgroup 16 condition for mmid subgroup optimization (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15565\"\u003e#15565\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/886b97a5d693550c2da470c091d9d27bf38398f8\"\u003e\u003ccode\u003e886b97a\u003c/code\u003e\u003c/a\u003e tests: Generate unique input values for count_equal (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15487\"\u003e#15487\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/111f8d06f0b9169059779a35f655b343242ccbb6\"\u003e\u003ccode\u003e111f8d0\u003c/code\u003e\u003c/a\u003e metal: fix regression when no metal devices are present (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15531\"\u003e#15531\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/5eff6ec9b1220b599a43b594b1110487ab6aca08\"\u003e\u003ccode\u003e5eff6ec\u003c/code\u003e\u003c/a\u003e CUDA: MoE helper in device code, better tile sizes (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15525\"\u003e#15525\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/dfd9b5f6c7586c88588f06a644c131bec071a0a1\"\u003e\u003ccode\u003edfd9b5f\u003c/code\u003e\u003c/a\u003e model-conversion : set pooling type to none in logits.cpp (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15564\"\u003e#15564\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/5a6bc6b1a6cb665a944426c2055794950e524bf5\"\u003e\u003ccode\u003e5a6bc6b\u003c/code\u003e\u003c/a\u003e model-conversion : add model card template for embeddings [no ci] (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15557\"\u003e#15557\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6b64f74b55628e4193f4fb00313f07dbd8556528\"\u003e\u003ccode\u003e6b64f74\u003c/code\u003e\u003c/a\u003e batched-bench : fix unified KV cache handling + pp timing (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15562\"\u003e#15562\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0d5a470223fc90b6b6807921d68011ff06ae7f9e\"\u003e\u003ccode\u003e0d5a470\u003c/code\u003e\u003c/a\u003e convert : update Ernie 4.5 dense architecture name (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15555\"\u003e#15555\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b0ba31f525a2ad7fb539660be0c8f088c9869f6a\"\u003e\u003ccode\u003eb0ba31f\u003c/code\u003e\u003c/a\u003e metal : add FA kernels for HS=40 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15559\"\u003e#15559\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/7da9fed0d6f1750a8783436f6f313b87e76e6378\"\u003e\u003ccode\u003e7da9fed\u003c/code\u003e\u003c/a\u003e convert : support interns1-mini (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15412\"\u003e#15412\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...4d917cd4f64cc37744e76d084659475819fb0728\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/21","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/21","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/21/packages"}},{"old_version":"`75c91de`","new_version":"`6d7f111`","update_type":null,"path":null,"pr_created_at":"2025-08-19T03:50:03.000Z","version_change":"`75c91de` → `6d7f111`","issue":{"uuid":"2755092012","node_id":"PR_kwDOOgWcXc6kN14s","number":20,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `6d7f111`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-08-26T09:38:09.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-08-19T03:50:03.000Z","updated_at":"2025-08-26T09:38:09.000Z","time_to_close":625686,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`6d7f111`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `6d7f111`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6d7f1117e3e3285d0c5c11b5ebb0439e27920082\"\u003e\u003ccode\u003e6d7f111\u003c/code\u003e\u003c/a\u003e codeowners : remove mmv.*\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/60212f1ead2dce9bf1ac69633a7069258ae604d8\"\u003e\u003ccode\u003e60212f1\u003c/code\u003e\u003c/a\u003e sync : ggml\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f0c541d315e97b297b3421c52ebde53340ee66b3\"\u003e\u003ccode\u003ef0c541d\u003c/code\u003e\u003c/a\u003e scripts : update sync scripts\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/baa9255a45105d2d3b4ec432af13b7a6eda3ff35\"\u003e\u003ccode\u003ebaa9255\u003c/code\u003e\u003c/a\u003e llama : merge conts and reshapes and remove unnecessary cont (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15380\"\u003e#15380\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/3007baf201e7ffcda17dbdb0335997fa50a6595b\"\u003e\u003ccode\u003e3007baf\u003c/code\u003e\u003c/a\u003e readme : update hot topics (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15397\"\u003e#15397\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/d1d82416006e7ff41780cb0e9b5f28d30a267497\"\u003e\u003ccode\u003ed1d8241\u003c/code\u003e\u003c/a\u003e server : fix incoming tasks not process in order (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15395\"\u003e#15395\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/618575c5825d7d4f170e686e772178d2aae148ae\"\u003e\u003ccode\u003e618575c\u003c/code\u003e\u003c/a\u003e Fix broken build: require updated pip to support --break-system-packages (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15\"\u003e#15\u003c/a\u003e...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f44f7931729022c57319a0124931120a169e0da9\"\u003e\u003ccode\u003ef44f793\u003c/code\u003e\u003c/a\u003e ggml-quants : fix make_qp_quants NANs and IQ1 assertion errors (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15379\"\u003e#15379\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ae532eac2c1df1d8edc3d2719145895b966de1bf\"\u003e\u003ccode\u003eae532ea\u003c/code\u003e\u003c/a\u003e vulkan: disable spirv-opt for bfloat16 shaders (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15352\"\u003e#15352\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/e5155e698645242d4f019267ecc40ea9bad81b09\"\u003e\u003ccode\u003ee5155e6\u003c/code\u003e\u003c/a\u003e server : export max observed n_past value (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15361\"\u003e#15361\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...6d7f1117e3e3285d0c5c11b5ebb0439e27920082\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/20","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/20","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/20/packages"}},{"old_version":"`75c91de`","new_version":"`be48528`","update_type":null,"path":null,"pr_created_at":"2025-08-12T00:03:47.000Z","version_change":"`75c91de` → `be48528`","issue":{"uuid":"2737209725","node_id":"PR_kwDOOgWcXc6jJoF9","number":18,"state":"open","title":"Bump vendored/llama.cpp from `75c91de` to `be48528`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":0,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-08-12T00:03:47.000Z","updated_at":"2025-08-12T00:03:48.000Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`be48528`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `be48528`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/be48528b068111304e4a0bb82c028558b5705f05\"\u003e\u003ccode\u003ebe48528\u003c/code\u003e\u003c/a\u003e CANN: Add broadcast for softmax and FA (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15208\"\u003e#15208\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/cf9e5648a7ae02f7728fefcbcfd2979d83c96e8f\"\u003e\u003ccode\u003ecf9e564\u003c/code\u003e\u003c/a\u003e mtmd : Fix MinicpmV model converter and clip to avoid using hardcode. (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14750\"\u003e#14750\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/fba5c0d680b555cbac563fef57b33bc5c9621e08\"\u003e\u003ccode\u003efba5c0d\u003c/code\u003e\u003c/a\u003e chat : hotfix gpt-oss jinja raising an exception (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15243\"\u003e#15243\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/53d0a1265826f40c5dbc01d06aeab9e14fcbd69b\"\u003e\u003ccode\u003e53d0a12\u003c/code\u003e\u003c/a\u003e server : allow specifying reasoning_format in HTTP request (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15238\"\u003e#15238\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/27093afe78912494073eb043fec93a007e49653c\"\u003e\u003ccode\u003e27093af\u003c/code\u003e\u003c/a\u003e readme : update infra list (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15234\"\u003e#15234\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/228f724d9ce6c56e8cec75bfdabab4dd013def7f\"\u003e\u003ccode\u003e228f724\u003c/code\u003e\u003c/a\u003e kv-cache : fix seq_rm with seq_id == -1 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15226\"\u003e#15226\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/cd3069dfcbeee8e0e96cffb93b0ebb9e595e273a\"\u003e\u003ccode\u003ecd3069d\u003c/code\u003e\u003c/a\u003e kv-cache : log (debug) all streams in find_slot (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15176\"\u003e#15176\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/50e81bdf5db563ab57a9a722b08f96fa8a76c927\"\u003e\u003ccode\u003e50e81bd\u003c/code\u003e\u003c/a\u003e convert : fix merge conflicts (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15229\"\u003e#15229\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/1ebbaddff2a44b0599df659175d3274bd5bbeb81\"\u003e\u003ccode\u003e1ebbadd\u003c/code\u003e\u003c/a\u003e perplexity : update comments/error msg to use decode [no ci] (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/15227\"\u003e#15227\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a3a7874272e5a060079658eb5cca4617b7f99062\"\u003e\u003ccode\u003ea3a7874\u003c/code\u003e\u003c/a\u003e convert : improve Mistral models integration (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14737\"\u003e#14737\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...be48528b068111304e4a0bb82c028558b5705f05\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/18","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/18","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/18/packages"}},{"old_version":"`75c91de`","new_version":"`6c9ee3b`","update_type":null,"path":null,"pr_created_at":"2025-07-21T22:22:01.000Z","version_change":"`75c91de` → `6c9ee3b`","issue":{"uuid":"2683315475","node_id":"PR_kwDOOgWcXc6f8CUT","number":16,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `6c9ee3b`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-07-21T22:42:43.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-07-21T22:22:01.000Z","updated_at":"2025-07-21T22:42:43.000Z","time_to_close":1242,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`6c9ee3b`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `6c9ee3b`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6c9ee3b17e19dcc82ab93d52ae46fdd0226d4777\"\u003e\u003ccode\u003e6c9ee3b\u003c/code\u003e\u003c/a\u003e opencl: add conv2d kernel (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14403\"\u003e#14403\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/cd465d823c378853f1f6570eebfb77f69c7e1d39\"\u003e\u003ccode\u003ecd465d8\u003c/code\u003e\u003c/a\u003e sycl: Fix im2col (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14797\"\u003e#14797\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/922042601b8a16877ccb1c2afaa2071f76734f10\"\u003e\u003ccode\u003e9220426\u003c/code\u003e\u003c/a\u003e kleidiai: add support for get_rows (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14676\"\u003e#14676\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/2ba1333b35e471b344974dde553db11bf1c2836f\"\u003e\u003ccode\u003e2ba1333\u003c/code\u003e\u003c/a\u003e docs : fix backends table in README.md (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14796\"\u003e#14796\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c2e058f1b4e799f1be085560c1bcef95b7b5ed02\"\u003e\u003ccode\u003ec2e058f\u003c/code\u003e\u003c/a\u003e vulkan/cuda: Fix im2col when KW!=KH (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14789\"\u003e#14789\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c82d48ec23fb8749c341d0838f6891fd5f6b6da0\"\u003e\u003ccode\u003ec82d48e\u003c/code\u003e\u003c/a\u003e llama : fix \u003ccode\u003e--reverse-prompt\u003c/code\u003e crashing issue (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14794\"\u003e#14794\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b4efd77f8ab407836ca73a5176f041650c5b2411\"\u003e\u003ccode\u003eb4efd77\u003c/code\u003e\u003c/a\u003e server : add parse_special option to /tokenize endpoint (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14783\"\u003e#14783\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/2be60cbc2707359241c2784f9d2e30d8fc7cdabb\"\u003e\u003ccode\u003e2be60cb\u003c/code\u003e\u003c/a\u003e docs : fix link for tools/perplexity in README.md (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14780\"\u003e#14780\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b526ad2668944a7b2b1721f60679153646313831\"\u003e\u003ccode\u003eb526ad2\u003c/code\u003e\u003c/a\u003e Documentation: Further revisions to the Vulkan section in build.md (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14785\"\u003e#14785\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/938b785764683c298e7805e712f8728489cc2f18\"\u003e\u003ccode\u003e938b785\u003c/code\u003e\u003c/a\u003e Clang-format: local files first + fix BinPacking (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14779\"\u003e#14779\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...6c9ee3b17e19dcc82ab93d52ae46fdd0226d4777\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/16","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/16","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/16/packages"}},{"old_version":"`75c91de`","new_version":"`bdca383`","update_type":null,"path":null,"pr_created_at":"2025-07-14T21:36:45.000Z","version_change":"`75c91de` → `bdca383`","issue":{"uuid":"2665470558","node_id":"PR_kwDOOgWcXc6e39pe","number":15,"state":"closed","title":"Bump vendored/llama.cpp from `75c91de` to `bdca383`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-07-21T22:22:03.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-07-14T21:36:45.000Z","updated_at":"2025-07-21T22:22:03.000Z","time_to_close":607518,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`75c91de`","new_version":"`bdca383`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `75c91de` to `bdca383`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/bdca38376f7e8dd928defe01ce6a16218a64b040\"\u003e\u003ccode\u003ebdca383\u003c/code\u003e\u003c/a\u003e sycl: Hotfix for non dnnl codepath (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14677\"\u003e#14677\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/55c509daf51d25bfaee9c8b8ce6abff103d4473b\"\u003e\u003ccode\u003e55c509d\u003c/code\u003e\u003c/a\u003e ggml : refactor llamafile_sgemm PPC code (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14673\"\u003e#14673\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/9c9e4fc6354fc811efa06a8eb7a86d3315cec9c8\"\u003e\u003ccode\u003e9c9e4fc\u003c/code\u003e\u003c/a\u003e llama-context: add ability to get logits (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14672\"\u003e#14672\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/494c5899cb76859f32ddd913534f2685fd684a3d\"\u003e\u003ccode\u003e494c589\u003c/code\u003e\u003c/a\u003e scripts: benchmark for HTTP server throughput (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14668\"\u003e#14668\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0f4c6ec0f1a9607ba67071f8a02c69b0afc2f91e\"\u003e\u003ccode\u003e0f4c6ec\u003c/code\u003e\u003c/a\u003e SYCL: use 1D kernel for set_rows (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14618\"\u003e#14618\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/65a3ebb0aa56d6c501466e0f950ad15105fd32d8\"\u003e\u003ccode\u003e65a3ebb\u003c/code\u003e\u003c/a\u003e sycl: Batched mulmat rework for oneDNN dispatch (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14617\"\u003e#14617\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0d9226763c82562186122f3b827fa3862864a19c\"\u003e\u003ccode\u003e0d92267\u003c/code\u003e\u003c/a\u003e llama : add jinja template for rwkv-world (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14665\"\u003e#14665\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/982e347255723fe6d02e60ee30cfdd0559c884c5\"\u003e\u003ccode\u003e982e347\u003c/code\u003e\u003c/a\u003e quantize : fix minor logic flaw in --tensor-type (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14572\"\u003e#14572\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/923e3ea2e3c96a0b4c208f53bc3bc90cdcdf13c0\"\u003e\u003ccode\u003e923e3ea\u003c/code\u003e\u003c/a\u003e cuda : add set rows for bf16 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14664\"\u003e#14664\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/e743cddb60dc3a8815b9de7dd7d5c491e61b2259\"\u003e\u003ccode\u003ee743cdd\u003c/code\u003e\u003c/a\u003e cuda : add ELU support (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14657\"\u003e#14657\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/75c91de6e955d5b8f3f28173f5040593e1964eb3...bdca38376f7e8dd928defe01ce6a16218a64b040\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/15","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/15","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/15/packages"}},{"old_version":"`0a5a3b5`","new_version":"`12f55c3`","update_type":null,"path":null,"pr_created_at":"2025-07-07T23:13:54.000Z","version_change":"`0a5a3b5` → `12f55c3`","issue":{"uuid":"3210489402","node_id":"PR_kwDOOgWcXc6d1qFG","number":14,"state":"open","title":"Bump vendored/llama.cpp from `0a5a3b5` to `12f55c3`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":2,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-07-07T23:13:54.000Z","updated_at":"2025-07-08T06:02:25.116Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`0a5a3b5`","new_version":"`12f55c3`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `0a5a3b5` to `12f55c3`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/12f55c302b35cfe900b84c5fe67c262026af9c44\"\u003e\u003ccode\u003e12f55c3\u003c/code\u003e\u003c/a\u003e llama : remove ggml_cont where possible (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14568\"\u003e#14568\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b9c3eefde1b67104bd993485ff38dd62abe9d70c\"\u003e\u003ccode\u003eb9c3eef\u003c/code\u003e\u003c/a\u003e CUDA: add bf16 and i32 to getrows (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14529\"\u003e#14529\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6491d6e4f1caf0ad2221865b4249ae6938a6308c\"\u003e\u003ccode\u003e6491d6e\u003c/code\u003e\u003c/a\u003e vulkan: increase LOAD_VEC_A to 8 (IQ1/IQ2) or 4 (IQ3) (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14485\"\u003e#14485\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/e592be15756ae546ba87bb5a5250b71248121971\"\u003e\u003ccode\u003ee592be1\u003c/code\u003e\u003c/a\u003e vulkan: fix rms_norm+mul fusion (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14545\"\u003e#14545\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a0374a67e2924f2e845cdc59dd67d9a44065a89c\"\u003e\u003ccode\u003ea0374a6\u003c/code\u003e\u003c/a\u003e vulkan: Handle updated FA dim2/3 definition (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14518\"\u003e#14518\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ddef99522d1ba74193b7394e803fab8db5c78bae\"\u003e\u003ccode\u003eddef995\u003c/code\u003e\u003c/a\u003e server : fix assistant prefilling when content is an array (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14360\"\u003e#14360\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/668168814601d2aef6161ce49ab186bc74177ae7\"\u003e\u003ccode\u003e6681688\u003c/code\u003e\u003c/a\u003e opencl: add GELU_ERF (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14476\"\u003e#14476\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/bac8bed248d15419137c5bc7f834582397baaebc\"\u003e\u003ccode\u003ebac8bed\u003c/code\u003e\u003c/a\u003e eval-callback : check for empty input (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14539\"\u003e#14539\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b81510a7b78f7c5a6f069c4f3d0e569b9d287913\"\u003e\u003ccode\u003eb81510a\u003c/code\u003e\u003c/a\u003e test-backend-ops: add support for specifying output format (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14368\"\u003e#14368\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ef797db357e44ecb7437fa9d22f4e1614104b342\"\u003e\u003ccode\u003eef797db\u003c/code\u003e\u003c/a\u003e metal : disable fast math in all quantize kernels (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14528\"\u003e#14528\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/0a5a3b5cdfd887cf0f8e09d9ff89dee130cfcdde...12f55c302b35cfe900b84c5fe67c262026af9c44\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/14","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/14","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/14/packages"}},{"old_version":"`0a5a3b5`","new_version":"`b9c3eef`","update_type":null,"path":null,"pr_created_at":"2025-07-07T23:12:15.000Z","version_change":"`0a5a3b5` → `b9c3eef`","issue":{"uuid":"2648085077","node_id":"PR_kwDOOgWcXc6d1pJV","number":13,"state":"open","title":"Bump vendored/llama.cpp from `0a5a3b5` to `b9c3eef`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":0,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-07-07T23:12:15.000Z","updated_at":"2025-07-07T23:12:16.000Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`0a5a3b5`","new_version":"`b9c3eef`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `0a5a3b5` to `b9c3eef`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b9c3eefde1b67104bd993485ff38dd62abe9d70c\"\u003e\u003ccode\u003eb9c3eef\u003c/code\u003e\u003c/a\u003e CUDA: add bf16 and i32 to getrows (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14529\"\u003e#14529\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6491d6e4f1caf0ad2221865b4249ae6938a6308c\"\u003e\u003ccode\u003e6491d6e\u003c/code\u003e\u003c/a\u003e vulkan: increase LOAD_VEC_A to 8 (IQ1/IQ2) or 4 (IQ3) (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14485\"\u003e#14485\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/e592be15756ae546ba87bb5a5250b71248121971\"\u003e\u003ccode\u003ee592be1\u003c/code\u003e\u003c/a\u003e vulkan: fix rms_norm+mul fusion (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14545\"\u003e#14545\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a0374a67e2924f2e845cdc59dd67d9a44065a89c\"\u003e\u003ccode\u003ea0374a6\u003c/code\u003e\u003c/a\u003e vulkan: Handle updated FA dim2/3 definition (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14518\"\u003e#14518\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ddef99522d1ba74193b7394e803fab8db5c78bae\"\u003e\u003ccode\u003eddef995\u003c/code\u003e\u003c/a\u003e server : fix assistant prefilling when content is an array (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14360\"\u003e#14360\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/668168814601d2aef6161ce49ab186bc74177ae7\"\u003e\u003ccode\u003e6681688\u003c/code\u003e\u003c/a\u003e opencl: add GELU_ERF (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14476\"\u003e#14476\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/bac8bed248d15419137c5bc7f834582397baaebc\"\u003e\u003ccode\u003ebac8bed\u003c/code\u003e\u003c/a\u003e eval-callback : check for empty input (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14539\"\u003e#14539\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b81510a7b78f7c5a6f069c4f3d0e569b9d287913\"\u003e\u003ccode\u003eb81510a\u003c/code\u003e\u003c/a\u003e test-backend-ops: add support for specifying output format (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14368\"\u003e#14368\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ef797db357e44ecb7437fa9d22f4e1614104b342\"\u003e\u003ccode\u003eef797db\u003c/code\u003e\u003c/a\u003e metal : disable fast math in all quantize kernels (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14528\"\u003e#14528\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/67d1ef23c68e1e09dc6d7423d1ef5fc0ea56ed5e\"\u003e\u003ccode\u003e67d1ef2\u003c/code\u003e\u003c/a\u003e batch : add optional for sequential equal split (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14511\"\u003e#14511\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/0a5a3b5cdfd887cf0f8e09d9ff89dee130cfcdde...b9c3eefde1b67104bd993485ff38dd62abe9d70c\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/13","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/13","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/13/packages"}},{"old_version":"`ce82bd0`","new_version":"`0a5a3b5`","update_type":null,"path":null,"pr_created_at":"2025-06-30T22:47:09.000Z","version_change":"`ce82bd0` → `0a5a3b5`","issue":{"uuid":"3190083416","node_id":"PR_kwDOOgWcXc6cx9gu","number":12,"state":"open","title":"Bump vendored/llama.cpp from `ce82bd0` to `0a5a3b5`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-06-30T22:47:09.000Z","updated_at":"2025-07-01T04:08:56.000Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`ce82bd0`","new_version":"`0a5a3b5`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"[//]: # (dependabot-start)\n⚠️  **Dependabot is rebasing this PR** ⚠️ \n\nRebasing might not happen immediately, so don't worry if this takes some time.\n\nNote: if you make any changes to this PR yourself, they will take precedence over the rebase.\n\n---\n\n[//]: # (dependabot-end)\n\nBumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `ce82bd0` to `0a5a3b5`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/0a5a3b5cdfd887cf0f8e09d9ff89dee130cfcdde\"\u003e\u003ccode\u003e0a5a3b5\u003c/code\u003e\u003c/a\u003e Add Conv2d for CPU (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14388\"\u003e#14388\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/745f11fed09ba2303f3f494efb12286d382608e5\"\u003e\u003ccode\u003e745f11f\u003c/code\u003e\u003c/a\u003e memory : correctly handle failure in apply() (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14438\"\u003e#14438\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/5dd942de5922a22ec8446a4ad2203738dbcb9389\"\u003e\u003ccode\u003e5dd942d\u003c/code\u003e\u003c/a\u003e metal : disable fast-math for some cpy kernels (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14460\"\u003e#14460\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/a7417f55945eb7c7a5ea6807e66564b8066a4e50\"\u003e\u003ccode\u003ea7417f5\u003c/code\u003e\u003c/a\u003e ggml-cpu: sycl: Re-enable exp f16 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14462\"\u003e#14462\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/eb3fa2913e0ace38912011fcb331624c679656f5\"\u003e\u003ccode\u003eeb3fa29\u003c/code\u003e\u003c/a\u003e test-backend-ops : disable llama test (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14461\"\u003e#14461\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c839a2da1a0d4275c3a98f70c3a7627487573a46\"\u003e\u003ccode\u003ec839a2d\u003c/code\u003e\u003c/a\u003e cmake : Remove redundant include path in CMakeLists.txt (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14452\"\u003e#14452\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/e9b6350e61d592634263a14b3d77ecbf6c1fb096\"\u003e\u003ccode\u003ee9b6350\u003c/code\u003e\u003c/a\u003e scripts : make the shell scripts cross-platform (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14341\"\u003e#14341\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/caf5681fcb47dfe9bafee94ef9aa8f669ac986c7\"\u003e\u003ccode\u003ecaf5681\u003c/code\u003e\u003c/a\u003e server : support jinja extra template kwargs (Qwen3 enable_thinking feature),...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/83790b0e7e09ab17238b16452a33053a71dbdfad\"\u003e\u003ccode\u003e83790b0\u003c/code\u003e\u003c/a\u003e server : fix appearance of the chats list context menu for Safari (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14322\"\u003e#14322\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f47c1d7106e49062279bcc57fc1077c0db61e278\"\u003e\u003ccode\u003ef47c1d7\u003c/code\u003e\u003c/a\u003e SYCL: disable faulty fp16 exp kernel (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14395\"\u003e#14395\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/ce82bd0117bd3598300b3a089d13d401b90279c7...0a5a3b5cdfd887cf0f8e09d9ff89dee130cfcdde\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/12","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/12","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/12/packages"}},{"old_version":"`6adc3c3`","new_version":"`ce82bd0`","update_type":null,"path":null,"pr_created_at":"2025-06-23T20:39:49.000Z","version_change":"`6adc3c3` → `ce82bd0`","issue":{"uuid":"2612741532","node_id":"PR_kwDOOgWcXc6bu0Wc","number":11,"state":"closed","title":"Bump vendored/llama.cpp from `6adc3c3` to `ce82bd0`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":3,"pull_request":true,"closed_at":"2025-06-27T15:25:16.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-06-23T20:39:49.000Z","updated_at":"2025-06-27T15:25:16.000Z","time_to_close":326727,"merged_at":"2025-06-27T15:25:16.000Z","merged_by":"dependabot[bot]","closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`6adc3c3`","new_version":"`ce82bd0`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `6adc3c3` to `ce82bd0`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ce82bd0117bd3598300b3a089d13d401b90279c7\"\u003e\u003ccode\u003ece82bd0\u003c/code\u003e\u003c/a\u003e ci: add workflow for relocatable cmake package (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14346\"\u003e#14346\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/bf2a99e3cb06a14dce2a586450d012dc31f922ae\"\u003e\u003ccode\u003ebf2a99e\u003c/code\u003e\u003c/a\u003e vulkan: update windows SDK in release.yml (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14344\"\u003e#14344\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/72c6bc3f3d0cf3bf160dddf1b803fed52bcbb0a3\"\u003e\u003ccode\u003e72c6bc3\u003c/code\u003e\u003c/a\u003e llama : better rwkv chat template and add missing \u003ccode\u003einputs.use_jinja\u003c/code\u003e setting ...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/defe2158dd5e250b4ef53994057a4ec03131a263\"\u003e\u003ccode\u003edefe215\u003c/code\u003e\u003c/a\u003e CUDA: mul_mat_v support for batch sizes \u0026gt; 1 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14262\"\u003e#14262\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/7b50d589a863c7631135c1226f6eab65cb406212\"\u003e\u003ccode\u003e7b50d58\u003c/code\u003e\u003c/a\u003e kv-cells : fix tracking of seq_pos (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14339\"\u003e#14339\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/3a9457df962b5883f2773f1c295e8c19df60d89f\"\u003e\u003ccode\u003e3a9457d\u003c/code\u003e\u003c/a\u003e vulkan: update windows SDK in CI (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14334\"\u003e#14334\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/fa4a9f2a1ccda2573189a9d4995bdf0bceb41156\"\u003e\u003ccode\u003efa4a9f2\u003c/code\u003e\u003c/a\u003e quantize : handle user-defined pruning of whole layers (blocks) (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13037\"\u003e#13037\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/238005c2dc67426cf678baa2d54c881701693288\"\u003e\u003ccode\u003e238005c\u003c/code\u003e\u003c/a\u003e gguf-py : fix SpecialVocab parsing when post_processor is null (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14330\"\u003e#14330\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/66aba7aca9a245d71f1ddf02c7a97223529752a8\"\u003e\u003ccode\u003e66aba7a\u003c/code\u003e\u003c/a\u003e run : avoid double tokenization (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14327\"\u003e#14327\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f1f5e82df6222dcbaca6396c0de44df259a1694f\"\u003e\u003ccode\u003ef1f5e82\u003c/code\u003e\u003c/a\u003e examples : fix is_first logic for tokenization (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14329\"\u003e#14329\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/6adc3c3ebc029af058ac950a8e2a825fdf18ecc6...ce82bd0117bd3598300b3a089d13d401b90279c7\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/11","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/11","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/11/packages"}},{"old_version":"`3637576`","new_version":"`7f4fbe5`","update_type":null,"path":null,"pr_created_at":"2025-06-09T18:31:50.000Z","version_change":"`3637576` → `7f4fbe5`","issue":{"uuid":"2579131426","node_id":"PR_kwDOOgWcXc6Zumwi","number":9,"state":"closed","title":"Bump vendored/llama.cpp from `3637576` to `7f4fbe5`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":"2025-06-16T02:43:12.000Z","author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-06-09T18:31:50.000Z","updated_at":"2025-06-16T02:43:12.000Z","time_to_close":547882,"merged_at":"2025-06-16T02:43:12.000Z","merged_by":"dependabot[bot]","closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`3637576`","new_version":"`7f4fbe5`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `3637576` to `7f4fbe5`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/7f4fbe5183b23b6b2e25fd1ccc5d1fa8bb010cb7\"\u003e\u003ccode\u003e7f4fbe5\u003c/code\u003e\u003c/a\u003e llama : allow building all tests on windows when not using shared libs (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13980\"\u003e#13980\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f470bc36bed4d836b9de5a483fa0dfaee176d6f5\"\u003e\u003ccode\u003ef470bc3\u003c/code\u003e\u003c/a\u003e ggml-cpu : split arch-specific implementations (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13892\"\u003e#13892\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/8f47e25f56e9792093b7497c68e9f80bab82ed19\"\u003e\u003ccode\u003e8f47e25\u003c/code\u003e\u003c/a\u003e cuda : fix device sync on buffer clear (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14033\"\u003e#14033\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/201b31dc2e6fef3de598280b53fd3b69420a4e49\"\u003e\u003ccode\u003e201b31d\u003c/code\u003e\u003c/a\u003e graph : fix geglu (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14077\"\u003e#14077\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/e21d2d4ae26d6c5daf1602a1061aa4f8e722ca57\"\u003e\u003ccode\u003ee21d2d4\u003c/code\u003e\u003c/a\u003e CANN: Simplify the environment variable setting(\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13104\"\u003e#13104\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/dc0623fddb661926e0998538895155e4f081ff09\"\u003e\u003ccode\u003edc0623f\u003c/code\u003e\u003c/a\u003e webui: fix sidebar being covered by main content (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14082\"\u003e#14082\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/87d34b381d5868e75586210ec17b5ef5deddc276\"\u003e\u003ccode\u003e87d34b3\u003c/code\u003e\u003c/a\u003e server : fix LRU check (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14079\"\u003e#14079\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/b460d16ae858c6624fd37aec316622a4060ca325\"\u003e\u003ccode\u003eb460d16\u003c/code\u003e\u003c/a\u003e sycl: Add reorder to Q6_K mmvq implementation (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13885\"\u003e#13885\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/91a8ee6a6f1f4c8547ff7b745ef95c6edc1d2af6\"\u003e\u003ccode\u003e91a8ee6\u003c/code\u003e\u003c/a\u003e add geglu activation function (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/14074\"\u003e#14074\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/056eb74534ffd3efc50a9b854156eb6876a52a44\"\u003e\u003ccode\u003e056eb74\u003c/code\u003e\u003c/a\u003e CANN: Enable labeler for Ascend NPU (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13914\"\u003e#13914\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/363757628848a27a435bbf22ff9476e9aeda5f40...7f4fbe5183b23b6b2e25fd1ccc5d1fa8bb010cb7\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/9","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/9","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/9/packages"}},{"old_version":"`6f180b9`","new_version":"`3637576`","update_type":null,"path":null,"pr_created_at":"2025-06-02T19:14:21.000Z","version_change":"`6f180b9` → `3637576`","issue":{"uuid":"3111325223","node_id":"PR_kwDOOgWcXc6Yr_1C","number":8,"state":"open","title":"Bump vendored/llama.cpp from `6f180b9` to `3637576`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-06-02T19:14:21.000Z","updated_at":"2025-06-03T04:33:56.000Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`6f180b9`","new_version":"`3637576`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `6f180b9` to `3637576`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/363757628848a27a435bbf22ff9476e9aeda5f40\"\u003e\u003ccode\u003e3637576\u003c/code\u003e\u003c/a\u003e server : disable speculative decoding for SWA models (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13970\"\u003e#13970\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/ea394d7ab1f8101716d48ce9421c94c71b93a00f\"\u003e\u003ccode\u003eea394d7\u003c/code\u003e\u003c/a\u003e metal : use F32 accumulators in FA kernels (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13975\"\u003e#13975\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/5582c49c3961269eca96822abfb87528e942dd07\"\u003e\u003ccode\u003e5582c49\u003c/code\u003e\u003c/a\u003e gemma : more consistent attention scaling for v2 and v3 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13951\"\u003e#13951\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c9bbc77931d223ed7e7cbcf1cb057bc02fd0db19\"\u003e\u003ccode\u003ec9bbc77\u003c/code\u003e\u003c/a\u003e \u003ccode\u003eserver\u003c/code\u003e: update deepseek reasoning format (pass reasoning_content as diffs) ...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/bfd322796cd838f906535ff3352624fc46338894\"\u003e\u003ccode\u003ebfd3227\u003c/code\u003e\u003c/a\u003e mtmd : fix memory leak in mtmd_helper_eval_chunk_single (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13961\"\u003e#13961\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/093e3f1feb16e25e58f7d61e01266c830dd424b8\"\u003e\u003ccode\u003e093e3f1\u003c/code\u003e\u003c/a\u003e cmake : Handle mixed-case 'Power' strings in POWER CPU detection (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13966\"\u003e#13966\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/663445b0deb21fb602176da030d4154197a4fca6\"\u003e\u003ccode\u003e663445b\u003c/code\u003e\u003c/a\u003e sycl: quantize and reorder the input to q8_1 when reorder is enabled (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13826\"\u003e#13826\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/7675c555a13c9f473249e59a54db35032ce8e0fc\"\u003e\u003ccode\u003e7675c55\u003c/code\u003e\u003c/a\u003e gguf: fix failure on version == 0 (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13956\"\u003e#13956\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/5e1c3aed4074480f63e914d6c44c93536ed1452a\"\u003e\u003ccode\u003e5e1c3ae\u003c/code\u003e\u003c/a\u003e convert : fix nomic-bert-moe mask token (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13757\"\u003e#13757\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/c496fe0b1da28618dd17bda8b0a6bf5004554080\"\u003e\u003ccode\u003ec496fe0\u003c/code\u003e\u003c/a\u003e convert : fix vocab padding code for bert models (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13954\"\u003e#13954\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/6f180b915c9ed9ec0c240b5dcd64644988fb5e82...363757628848a27a435bbf22ff9476e9aeda5f40\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/8","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/8","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/8/packages"}},{"old_version":"`6a2bc8b`","new_version":"`6f180b9`","update_type":null,"path":null,"pr_created_at":"2025-05-26T18:48:01.000Z","version_change":"`6a2bc8b` → `6f180b9`","issue":{"uuid":"3091869130","node_id":"PR_kwDOOgWcXc6XrJEs","number":7,"state":"open","title":"Bump vendored/llama.cpp from `6a2bc8b` to `6f180b9`","user":"dependabot[bot]","labels":["dependencies","submodules"],"assignees":[],"locked":false,"comments_count":1,"pull_request":true,"closed_at":null,"author_association":"CONTRIBUTOR","state_reason":null,"created_at":"2025-05-26T18:48:01.000Z","updated_at":"2025-05-28T03:06:19.000Z","time_to_close":null,"merged_at":null,"merged_by":null,"closed_by":null,"dependency_metadata":{"prefix":"Bump","packages":[{"name":"vendored/llama.cpp","old_version":"`6a2bc8b`","new_version":"`6f180b9`","repository_url":"https://github.com/ggml-org/llama.cpp"}],"path":null,"ecosystem":"submodules"},"body":"Bumps [vendored/llama.cpp](https://github.com/ggml-org/llama.cpp) from `6a2bc8b` to `6f180b9`.\n\u003cdetails\u003e\n\u003csummary\u003eCommits\u003c/summary\u003e\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/6f180b915c9ed9ec0c240b5dcd64644988fb5e82\"\u003e\u003ccode\u003e6f180b9\u003c/code\u003e\u003c/a\u003e SYCL: Add non contiguous support in RMS_NORM and NORM kernels (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13611\"\u003e#13611\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/03f582ae8fccecff225c30a2802461b44761e822\"\u003e\u003ccode\u003e03f582a\u003c/code\u003e\u003c/a\u003e server: fix streaming crashes (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13786\"\u003e#13786\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/88c125f2acce0e25e5fc8481ab0681415fc64a10\"\u003e\u003ccode\u003e88c125f\u003c/code\u003e\u003c/a\u003e examples/training: Fix file name in README (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13803\"\u003e#13803\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/d74e94c1b3ac02db01e47008e1f8d371029d81ac\"\u003e\u003ccode\u003ed74e94c\u003c/code\u003e\u003c/a\u003e \u003ccode\u003eserver\u003c/code\u003e: fix format of streamed tool call deltas (diff name, fix id location...\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/f13847cfb560d92492b480cca2c5d6aa9473cde3\"\u003e\u003ccode\u003ef13847c\u003c/code\u003e\u003c/a\u003e server: fix regression on streamed non-chat completion w/ stops (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13785\"\u003e#13785\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/79c137f77677b3c8ee3c60a7da033721b938399a\"\u003e\u003ccode\u003e79c137f\u003c/code\u003e\u003c/a\u003e examples : allow extracting embeddings from decoder contexts (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13797\"\u003e#13797\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/22229314fc46b2f741bb21b12cde71f6c6a60b52\"\u003e\u003ccode\u003e2222931\u003c/code\u003e\u003c/a\u003e llama : clarify deprecation message (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13794\"\u003e#13794\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/9012eb9b454a82eaa4cd77ae904c0ea391e4db42\"\u003e\u003ccode\u003e9012eb9\u003c/code\u003e\u003c/a\u003e sycl: Add more debug prints (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13640\"\u003e#13640\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/fef693dc6b959a8e8ba11558fbeaad0b264dd457\"\u003e\u003ccode\u003efef693d\u003c/code\u003e\u003c/a\u003e vulkan: mark IM2COL as supporting non-contig (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13783\"\u003e#13783\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003e\u003ca href=\"https://github.com/ggml-org/llama.cpp/commit/2d38b6e4004fb1c341723e657fb1e71a4d3fb473\"\u003e\u003ccode\u003e2d38b6e\u003c/code\u003e\u003c/a\u003e CANN: Add the basic supports of Flash Attention kernel (\u003ca href=\"https://redirect.github.com/ggml-org/llama.cpp/issues/13627\"\u003e#13627\u003c/a\u003e)\u003c/li\u003e\n\u003cli\u003eAdditional commits viewable in \u003ca href=\"https://github.com/ggml-org/llama.cpp/compare/6a2bc8bfb7cd502e5ebc72e36c97a6f848c21c2c...6f180b915c9ed9ec0c240b5dcd64644988fb5e82\"\u003ecompare view\u003c/a\u003e\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/details\u003e\n\u003cbr /\u003e\n\n\nDependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`.\n\n[//]: # (dependabot-automerge-start)\n[//]: # (dependabot-automerge-end)\n\n---\n\n\u003cdetails\u003e\n\u003csummary\u003eDependabot commands and options\u003c/summary\u003e\n\u003cbr /\u003e\n\nYou can trigger Dependabot actions by commenting on this PR:\n- `@dependabot rebase` will rebase this PR\n- `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it\n- `@dependabot merge` will merge this PR after your CI passes on it\n- `@dependabot squash and merge` will squash and merge this PR after your CI passes on it\n- `@dependabot cancel merge` will cancel a previously requested merge and block automerging\n- `@dependabot reopen` will reopen this PR if it is closed\n- `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually\n- `@dependabot show \u003cdependency name\u003e ignore conditions` will show all of the ignore conditions of the specified dependency\n- `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)\n- `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)\n\n\n\u003c/details\u003e","html_url":"https://github.com/TheCBaH/ocaml-llama.cpp/pull/7","url":"https://dependabot.ecosyste.ms/api/v1/hosts/GitHub/repositories/TheCBaH%2Focaml-llama.cpp/issues/7","packages_url":"https://dependabot.ecosyste.ms/api/v1/issues/7/packages"}}]}