{"payload":{"feedbackUrl":"https://github.com/orgs/community/discussions/53140","repo":{"id":612354784,"defaultBranch":"master","name":"llama.cpp","ownerLogin":"ggerganov","currentUserCanPush":false,"isFork":false,"isEmpty":false,"createdAt":"2023-03-10T18:58:00.000Z","ownerAvatar":"https://avatars.githubusercontent.com/u/1991296?v=4","public":true,"private":false,"isOrgOwned":false},"refInfo":{"name":"","listCacheKey":"v0:1716051524.0","currentOid":""},"activityList":{"items":[{"before":"511182eabb36f6ec9776e2b3c4d7e16d93d0ac0d","after":"059031b8c40e1f4ba60586842c5b1ed3ddf61842","ref":"refs/heads/master","pushedAt":"2024-05-18T15:55:54.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"ggerganov","name":"Georgi Gerganov","path":"/ggerganov","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/1991296?s=80&v=4"},"commit":{"message":"ci : re-enable sanitizer runs (#7358)\n\n* Revert \"ci : temporary disable sanitizer builds (#6128)\"\r\n\r\nThis reverts commit 4f6d1337ca5a409dc74aca8c479b7c34408a69c0.\r\n\r\n* ci : trigger","shortMessageHtmlLink":"ci : re-enable sanitizer runs (#7358)"}},{"before":"133d99c59980139f5bb75922c8b5fca67d7ba9b8","after":"511182eabb36f6ec9776e2b3c4d7e16d93d0ac0d","ref":"refs/heads/master","pushedAt":"2024-05-18T10:40:39.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"mofosyne","name":"Brian","path":"/mofosyne","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/827793?s=80&v=4"},"commit":{"message":"android : use \"ci-android\" branch for CI (#7341)\n\n* android : use \"ci-android\" branch for CI\r\n\r\n* ggml : disable SIMD exp and silu for 32-bit ARM\r\n\r\nggml-ci\r\n\r\n* android : do not fetch, use add_subdirectory instead\r\n\r\n* cmake : provide binary dir","shortMessageHtmlLink":"android : use \"ci-android\" branch for CI (#7341)"}},{"before":"cb42c294279bc4a0a4e926a7b5a5568049f12fa7","after":"133d99c59980139f5bb75922c8b5fca67d7ba9b8","ref":"refs/heads/master","pushedAt":"2024-05-18T10:36:25.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"JohannesGaessler","name":"Johannes Gäßler","path":"/JohannesGaessler","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18492268?s=80&v=4"},"commit":{"message":"CUDA: deduplicate FlashAttention code (#7352)","shortMessageHtmlLink":"CUDA: deduplicate FlashAttention code (#7352)"}},{"before":"d233b507cd19fcc2d8d8963ecc6a3eb7a33f2ecc","after":"cb42c294279bc4a0a4e926a7b5a5568049f12fa7","ref":"refs/heads/master","pushedAt":"2024-05-18T09:10:47.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"JohannesGaessler","name":"Johannes Gäßler","path":"/JohannesGaessler","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18492268?s=80&v=4"},"commit":{"message":"server: correct --threads documentation [no ci] (#7362)","shortMessageHtmlLink":"server: correct --threads documentation [no ci] (#7362)"}},{"before":"052e9d59bc7cfbe0e3cca3ee10fd2e303d4ae13e","after":"1781015ba82afb1b7c4806b5c5016ff3b9232f6f","ref":"refs/heads/gg/ci-enable-sanitizers","pushedAt":"2024-05-18T08:41:31.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"ggerganov","name":"Georgi Gerganov","path":"/ggerganov","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/1991296?s=80&v=4"},"commit":{"message":"ci : trigger","shortMessageHtmlLink":"ci : trigger"}},{"before":null,"after":"8dbde1f033ac28c782dc4de0d83926f648970372","ref":"refs/heads/0cc4m/vulkan-embedding-fix","pushedAt":"2024-05-18T08:19:56.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"0cc4m","name":null,"path":"/0cc4m","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/11707594?s=80&v=4"},"commit":{"message":"Remove deprecated tensor->backend uses","shortMessageHtmlLink":"Remove deprecated tensor->backend uses"}},{"before":"0f98acfac6cc561dc57586bfff778405e42b576b","after":"d233b507cd19fcc2d8d8963ecc6a3eb7a33f2ecc","ref":"refs/heads/master","pushedAt":"2024-05-18T08:05:17.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"JohannesGaessler","name":"Johannes Gäßler","path":"/JohannesGaessler","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18492268?s=80&v=4"},"commit":{"message":"cuda : add half2 __shfl_xor() for ROCm 5.5 (#7263)","shortMessageHtmlLink":"cuda : add half2 __shfl_xor() for ROCm 5.5 (#7263)"}},{"before":"ca57e0f35e33f714b9a6c2c4482b87bfe059c819","after":"0f98acfac6cc561dc57586bfff778405e42b576b","ref":"refs/heads/master","pushedAt":"2024-05-18T08:04:55.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"ggerganov","name":"Georgi Gerganov","path":"/ggerganov","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/1991296?s=80&v=4"},"commit":{"message":"llama : add support for larger Granite Code Models (20B, 34B) (#7324)\n\nTie the weights for ARCH_STARCODER to support the larger Granite code models.\r\nPartially addresses ggerganov/issues/7116\r\n\r\nThere still remains to be a few things to fix.\r\nCurrently requires `--override-kv tokenizer.ggml.add_bos_token=bool:false`","shortMessageHtmlLink":"llama : add support for larger Granite Code Models (20B, 34B) (#7324)"}},{"before":"c1b295eea5c49887a066559527a74e8b94fe9db0","after":"ca57e0f35e33f714b9a6c2c4482b87bfe059c819","ref":"refs/heads/master","pushedAt":"2024-05-18T07:57:08.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"ggerganov","name":"Georgi Gerganov","path":"/ggerganov","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/1991296?s=80&v=4"},"commit":{"message":"perplexity : ndot progress and show stats with < 100 tasks (#7348)\n\nFix floating point error with ndot printing, allow end stats on lower task numbers if multiple-choice tasks.","shortMessageHtmlLink":"perplexity : ndot progress and show stats with < 100 tasks (#7348)"}},{"before":"99d1e7eb8a97090f8478fc5596999a6df8739063","after":"007f2ece0a3e6ed45b02524c47540384e0fce0bc","ref":"refs/heads/ci-android","pushedAt":"2024-05-18T07:50:57.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"ggerganov","name":"Georgi Gerganov","path":"/ggerganov","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/1991296?s=80&v=4"},"commit":{"message":"cmake : provide binary dir","shortMessageHtmlLink":"cmake : provide binary dir"}},{"before":null,"after":"052e9d59bc7cfbe0e3cca3ee10fd2e303d4ae13e","ref":"refs/heads/gg/ci-enable-sanitizers","pushedAt":"2024-05-18T07:26:30.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"ggerganov","name":"Georgi Gerganov","path":"/ggerganov","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/1991296?s=80&v=4"},"commit":{"message":"Revert \"ci : temporary disable sanitizer builds (#6128)\"\n\nThis reverts commit 4f6d1337ca5a409dc74aca8c479b7c34408a69c0.","shortMessageHtmlLink":"Revert \"ci : temporary disable sanitizer builds (#6128)\""}},{"before":"2117b3038050c07a443770b7f086c9681030ddbe","after":"99d1e7eb8a97090f8478fc5596999a6df8739063","ref":"refs/heads/ci-android","pushedAt":"2024-05-18T06:33:27.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"ggerganov","name":"Georgi Gerganov","path":"/ggerganov","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/1991296?s=80&v=4"},"commit":{"message":"android : do not fetch, use add_subdirectory instead","shortMessageHtmlLink":"android : do not fetch, use add_subdirectory instead"}},{"before":"9c5d3fcffb3350a8e959281540f482e6780b5bbc","after":null,"ref":"refs/heads/0cc4m/soft-max-fix","pushedAt":"2024-05-18T06:11:03.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"0cc4m","name":null,"path":"/0cc4m","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/11707594?s=80&v=4"}},{"before":"de731963441ff128248259e1b99573d75264d210","after":"c1b295eea5c49887a066559527a74e8b94fe9db0","ref":"refs/heads/master","pushedAt":"2024-05-18T06:10:58.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"0cc4m","name":null,"path":"/0cc4m","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/11707594?s=80&v=4"},"commit":{"message":"Update and fix Vulkan soft_max and argsort implementations (#7237)\n\n* Update and fix Vulkan softmax implementation\r\n\r\n* Update and fix Vulkan argsort implementation","shortMessageHtmlLink":"Update and fix Vulkan soft_max and argsort implementations (#7237)"}},{"before":"b49a13dd2fa9c94c2c19a8c248bb7fa45499f9a8","after":"de731963441ff128248259e1b99573d75264d210","ref":"refs/heads/master","pushedAt":"2024-05-18T06:04:23.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"mofosyne","name":"Brian","path":"/mofosyne","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/827793?s=80&v=4"},"commit":{"message":"github-actions-labeler: initial commit (#7330)\n\n* github-actions-labeler: initial commit [no ci]\r\n\r\n* github actions: remove priority auto labeling [no ci]","shortMessageHtmlLink":"github-actions-labeler: initial commit (#7330)"}},{"before":"d12c57b559c326fc80266a8f10536e5d90564cfc","after":null,"ref":"refs/heads/gg/fix-spm-vocab-convert","pushedAt":"2024-05-18T05:46:22.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"ggerganov","name":"Georgi Gerganov","path":"/ggerganov","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/1991296?s=80&v=4"}},{"before":"05834841dcb4f922983ea976539c70472272df9a","after":"b49a13dd2fa9c94c2c19a8c248bb7fa45499f9a8","ref":"refs/heads/master","pushedAt":"2024-05-18T05:46:21.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"ggerganov","name":"Georgi Gerganov","path":"/ggerganov","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/1991296?s=80&v=4"},"commit":{"message":"convert : fix set_vocab_sentencepiece (#6866)\n\n* convert : fix set_vocab_sentencepiece\r\n\r\n* Update convert-hf-to-gguf.py","shortMessageHtmlLink":"convert : fix set_vocab_sentencepiece (#6866)"}},{"before":"f07e570c032b17c1a0a5a8ca6da7339929e83ea3","after":null,"ref":"refs/heads/sl/fix-quant-near-zero","pushedAt":"2024-05-18T00:39:57.000Z","pushType":"branch_deletion","commitsCount":0,"pusher":{"login":"slaren","name":null,"path":"/slaren","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/2141330?s=80&v=4"}},{"before":"ef277de2add255a08b2b909ebfbf70364d1f4dc4","after":"05834841dcb4f922983ea976539c70472272df9a","ref":"refs/heads/master","pushedAt":"2024-05-18T00:39:54.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"slaren","name":null,"path":"/slaren","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/2141330?s=80&v=4"},"commit":{"message":"ggml : fix quants nans when all the group weights are very close to zero (#7313)","shortMessageHtmlLink":"ggml : fix quants nans when all the group weights are very close to z…"}},{"before":"b43272afa29a64dcb8bcf26a96a05bac40792b92","after":"ef277de2add255a08b2b909ebfbf70364d1f4dc4","ref":"refs/heads/master","pushedAt":"2024-05-18T00:39:25.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"slaren","name":null,"path":"/slaren","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/2141330?s=80&v=4"},"commit":{"message":"cmake : fix typo in AMDGPU_TARGETS (#7356)","shortMessageHtmlLink":"cmake : fix typo in AMDGPU_TARGETS (#7356)"}},{"before":"61e8a0adac3cf340bc0e90e25ec498e0bb4cc7fd","after":"f07e570c032b17c1a0a5a8ca6da7339929e83ea3","ref":"refs/heads/sl/fix-quant-near-zero","pushedAt":"2024-05-17T23:29:06.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"slaren","name":null,"path":"/slaren","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/2141330?s=80&v=4"},"commit":{"message":"use higher eps only for the quants that need it\n\nggml-ci","shortMessageHtmlLink":"use higher eps only for the quants that need it"}},{"before":"6b41894a025f63a79593a5c79b6423542f6cc54c","after":"61e8a0adac3cf340bc0e90e25ec498e0bb4cc7fd","ref":"refs/heads/sl/fix-quant-near-zero","pushedAt":"2024-05-17T23:25:58.000Z","pushType":"force_push","commitsCount":0,"pusher":{"login":"slaren","name":null,"path":"/slaren","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/2141330?s=80&v=4"},"commit":{"message":"use higher eps only for the quants that need it\n\nggml-ci","shortMessageHtmlLink":"use higher eps only for the quants that need it"}},{"before":"f59edeeae9780a4384c61d926ce46f73072224f5","after":"6b41894a025f63a79593a5c79b6423542f6cc54c","ref":"refs/heads/sl/fix-quant-near-zero","pushedAt":"2024-05-17T23:16:01.000Z","pushType":"push","commitsCount":1,"pusher":{"login":"slaren","name":null,"path":"/slaren","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/2141330?s=80&v=4"},"commit":{"message":"use higher eps only for the quants that need it\n\nggml-ci","shortMessageHtmlLink":"use higher eps only for the quants that need it"}},{"before":"0fc1e820a9900a3dd08ddd3c6abe6604c53b689b","after":"b43272afa29a64dcb8bcf26a96a05bac40792b92","ref":"refs/heads/master","pushedAt":"2024-05-17T23:09:13.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"jaime-m-p","name":null,"path":"/jaime-m-p","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/167997752?s=80&v=4"},"commit":{"message":"Unicode codepoint flags for custom regexs (#7245)\n\n* Replace CODEPOINT_TYPE_* with codepoint_flags\r\n* Update and bugfix brute force random test\r\n* Deterministic brute force random test\r\n* Unicode normalization NFD\r\n* Get rid of BOM","shortMessageHtmlLink":"Unicode codepoint flags for custom regexs (#7245)"}},{"before":"82ca83db3c8d45df559c03a4225b6eb34808a2db","after":"0fc1e820a9900a3dd08ddd3c6abe6604c53b689b","ref":"refs/heads/master","pushedAt":"2024-05-17T16:54:52.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"JohannesGaessler","name":"Johannes Gäßler","path":"/JohannesGaessler","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/18492268?s=80&v=4"},"commit":{"message":"CUDA: faster large batch FA without tensor cores (#7314)","shortMessageHtmlLink":"CUDA: faster large batch FA without tensor cores (#7314)"}},{"before":"f4bd8b3d260bb09491ba63c77ab7012b744362ef","after":"82ca83db3c8d45df559c03a4225b6eb34808a2db","ref":"refs/heads/master","pushedAt":"2024-05-17T15:03:03.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"slaren","name":null,"path":"/slaren","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/2141330?s=80&v=4"},"commit":{"message":"ROCm: use native CMake HIP support (#5966)\n\nSupercedes #4024 and #4813.\r\n\r\nCMake's native HIP support has become the\r\nrecommended way to add HIP code into a project (see\r\n[here](https://rocm.docs.amd.com/en/docs-6.0.0/conceptual/cmake-packages.html#using-hip-in-cmake)).\r\nThis PR makes the following changes:\r\n\r\n1. The environment variable `HIPCXX` or CMake option\r\n`CMAKE_HIP_COMPILER` should be used to specify the HIP\r\ncompiler. Notably this shouldn't be `hipcc`, but ROCm's clang,\r\nwhich usually resides in `$ROCM_PATH/llvm/bin/clang`. Previously\r\nthis was control by `CMAKE_C_COMPILER` and `CMAKE_CXX_COMPILER`.\r\nNote that since native CMake HIP support is not yet available on\r\nWindows, on Windows we fall back to the old behavior.\r\n\r\n2. CMake option `CMAKE_HIP_ARCHITECTURES` is used to control the\r\nGPU architectures to build for. Previously this was controled by\r\n`GPU_TARGETS`.\r\n\r\n3. Updated the Nix recipe to account for these new changes.\r\n\r\n4. The GPU targets to build against in the Nix recipe is now\r\nconsistent with the supported GPU targets in nixpkgs.\r\n\r\n5. Added CI checks for HIP on both Linux and Windows. On Linux, we test\r\nboth the new and old behavior.\r\n\r\nThe most important part about this PR is the separation of the\r\nHIP compiler and the C/C++ compiler. This allows users to choose\r\na different C/C++ compiler if desired, compared to the current\r\nsituation where when building for ROCm support, everything must be\r\ncompiled with ROCm's clang.\r\n\r\n~~Makefile is unchanged. Please let me know if we want to be\r\nconsistent on variables' naming because Makefile still uses\r\n`GPU_TARGETS` to control architectures to build for, but I feel\r\nlike setting `CMAKE_HIP_ARCHITECTURES` is a bit awkward when you're\r\ncalling `make`.~~ Makefile used `GPU_TARGETS` but the README says\r\nto use `AMDGPU_TARGETS`. For consistency with CMake, all usage of\r\n`GPU_TARGETS` in Makefile has been updated to `AMDGPU_TARGETS`.\r\n\r\nThanks to the suggestion of @jin-eld, to maintain backwards\r\ncompatibility (and not break too many downstream users' builds), if\r\n`CMAKE_CXX_COMPILER` ends with `hipcc`, then we still compile using\r\nthe original behavior and emit a warning that recommends switching\r\nto the new HIP support. Similarly, if `AMDGPU_TARGETS` is set but\r\n`CMAKE_HIP_ARCHITECTURES` is not, then we forward `AMDGPU_TARGETS`\r\nto `CMAKE_HIP_ARCHITECTURES` to ease the transition to the new\r\nHIP support.\r\n\r\nSigned-off-by: Gavin Zhao ","shortMessageHtmlLink":"ROCm: use native CMake HIP support (#5966)"}},{"before":"51e9d02599336e62948d29f1d6c05addeb921ac2","after":"f4bd8b3d260bb09491ba63c77ab7012b744362ef","ref":"refs/heads/master","pushedAt":"2024-05-17T14:25:44.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"rgerganov","name":"Radoslav Gerganov","path":"/rgerganov","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/271616?s=80&v=4"},"commit":{"message":"rpc : set SO_REUSEADDR for the server socket (#7320)\n\nref: #7293","shortMessageHtmlLink":"rpc : set SO_REUSEADDR for the server socket (#7320)"}},{"before":null,"after":"2117b3038050c07a443770b7f086c9681030ddbe","ref":"refs/heads/ci-android","pushedAt":"2024-05-17T12:48:22.000Z","pushType":"branch_creation","commitsCount":0,"pusher":{"login":"ggerganov","name":"Georgi Gerganov","path":"/ggerganov","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/1991296?s=80&v=4"},"commit":{"message":"ggml : disable SIMD exp and silu for 32-bit ARM\n\nggml-ci","shortMessageHtmlLink":"ggml : disable SIMD exp and silu for 32-bit ARM"}},{"before":"d273c1402b25086fd91aef2467ac13f2e49fa0ea","after":"51e9d02599336e62948d29f1d6c05addeb921ac2","ref":"refs/heads/master","pushedAt":"2024-05-17T12:40:14.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"mofosyne","name":"Brian","path":"/mofosyne","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/827793?s=80&v=4"},"commit":{"message":"Added a single test function script and fix debug-test.sh to be more robust (#7279)\n\n* run-single-test.sh: added a single test function script and fix debug-test.sh to be more robust\r\n\r\n* debug-test.sh: combined execute and gdb test mode via -g flag\r\n\r\n* debug-test.sh: refactor\r\n\r\n* debug-test: refactor for clarity\r\n\r\n* debug-test.sh: comment style changes\r\n\r\n* debug-test.sh: fix gdb","shortMessageHtmlLink":"Added a single test function script and fix debug-test.sh to be more …"}},{"before":"27b040691cbe45314147c2745e891a38e9c048d4","after":"d273c1402b25086fd91aef2467ac13f2e49fa0ea","ref":"refs/heads/master","pushedAt":"2024-05-17T12:11:45.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"ggerganov","name":"Georgi Gerganov","path":"/ggerganov","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/1991296?s=80&v=4"},"commit":{"message":"py : convert-hf-to-gguf-update improvements (#7340)\n\n* convert-hf-to-gguf-update: automate updating\r\n\r\n* convert-hf-to-gguf-update: improve download\r\n\r\n* share requests session for performance\r\n* create directories only when needed, don't skip downloads when empty directory encountered\r\n* be more graceful about errors","shortMessageHtmlLink":"py : convert-hf-to-gguf-update improvements (#7340)"}}],"hasNextPage":true,"hasPreviousPage":false,"activityType":"all","actor":null,"timePeriod":"all","sort":"DESC","perPage":30,"cursor":"djE6ks8AAAAETa9ClgA","startCursor":null,"endCursor":null}},"title":"Activity · ggerganov/llama.cpp"}