tag:github.com,2008:https://github.com/virajwad/llama.cpp/releases 2026-01-12T20:43:02Z tag:github.com,2008:Repository/1095211711/b7717 2026-01-12T20:43:02Z <p>ci : remove libcurl in releases (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/18775">ggml-org#18775</a>)</p> <p>Signed-off-by: Adrien Gallouët &lt;angt@huggingface.co&gt;</p> angt tag:github.com,2008:Repository/1095211711/b7648 2026-01-06T16:51:08Z <p>arg: use CSV escape style for multiple-value args (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/18643">ggml-org#18643</a>)</p> <p>* arg: use CSV escape style for multiple-value args</p> <p>* add test</p> ngxson tag:github.com,2008:Repository/1095211711/b7035 2025-11-12T13:17:24Z ngxson tag:github.com,2008:Repository/1095211711/b7034 2025-11-12T12:52:19Z <p>ggml-cpu: handle 3d tensors in repack mat_mul (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/17030">ggml-org#17030</a>)</p> <p>* ggml-cpu: handle 3d tensors in repack mul_mat</p> <p>* Removed unnecessary branch, removed need for &lt;algorithm&gt;</p> <p>* Fixed dst_ptr pointer in chunk + clang_format</p> <p>* GGML_ASSERT to check wdata within bounds</p> <p>* Accidental ggml.h inclusion</p> <p>* Improved GGML_ASSERT on wdata boundaries</p> Alcpz tag:github.com,2008:Repository/1095211711/b7033 2025-11-12T12:48:30Z angt tag:github.com,2008:Repository/1095211711/b7032 2025-11-12T11:32:50Z <p>cmake : move OpenSSL linking to vendor/cpp-httplib (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/17177">ggml-org#17177</a>)</p> <p>* cmake : move OpenSSL linking to vendor/cpp-httplib</p> <p>Signed-off-by: Adrien Gallouët &lt;angt@huggingface.co&gt;</p> <p>* bring back httplib 0.27.0</p> <p>* add -DLLAMA_HTTPLIB</p> <p>* update cmake config for visionos</p> <p>---------</p> <p>Signed-off-by: Adrien Gallouët &lt;angt@huggingface.co&gt; <br />Co-authored-by: Xuan Son Nguyen &lt;son@huggingface.co&gt;</p> angt tag:github.com,2008:Repository/1095211711/b7031 2025-11-12T07:11:42Z <p>CANN: Add L2_NORM op support (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/16856">ggml-org#16856</a>)</p> <p>* update L2_NORM op support</p> <p>* update L2_NORM op support</p> <p>* remove extra whitespace</p> TecJesh tag:github.com,2008:Repository/1095211711/b7030 2025-11-12T06:44:29Z <p>[SYCL]fix ci crash about SSM_CONV (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/17169">ggml-org#17169</a>)</p> <p>* fix ci crash</p> <p>* Update ggml-sycl.cpp</p> <p>* Update ggml/src/ggml-sycl/ggml-sycl.cpp</p> <p>Co-authored-by: Sigbjørn Skjæret &lt;sigbjorn.skjaeret@scala.com&gt;</p> <p>---------</p> <p>Co-authored-by: Zhang Jianyu &lt;zhang.jianyu@outlook.com&gt; <br />Co-authored-by: Sigbjørn Skjæret &lt;sigbjorn.skjaeret@scala.com&gt;</p> NeoZhangJianyu tag:github.com,2008:Repository/1095211711/b7028 2025-11-11T23:25:04Z <p>hexagon: various Op fixes (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/17135">ggml-org#17135</a>)</p> <p>* hexagon: explicitly check for ops with zero nrows</p> <p>llm_graph_context::build_inp_out_ids() can generate tensors with zero nrows. <br />Somehow other backends seems to handle this without obvious explicit checks. <br />In the hexagon case we need to check explicitly and skip them.</p> <p>* hexagon: introduce fastdiv, fix test-backend-ops for ADD/SUB/MUL</p> <p>Co-authored-by: chraac &lt;chraac@gmail.com&gt;</p> <p>* hexagon: use fastdiv in ADD_ID</p> <p>* hexagon: use ggml_op_is_empty and ggml_is_empty to check for NOPs</p> <p>---------</p> <p>Co-authored-by: chraac &lt;chraac@gmail.com&gt;</p> max-krasnyansky tag:github.com,2008:Repository/1095211711/b7027 2025-11-11T18:53:30Z netrunnereve