{"full_name":"ggml-org/ggml","owner":"ggml-org","name":"ggml","shard":"1","url":"https://github.com/ggml-org/ggml","description":"Tensor library for machine learning","homepage":"https://github.com/ggml-org","topics":["machine-learning"],"tag_pages":["machine-learning"],"language":"C++","license":"MIT","avatar":"https://avatars.githubusercontent.com/u/134263123?v=4","created_at":"2022-09-18T17:07:19Z","pushed_at":"2026-11-13T08:34:20Z","metrics":{"stars":16035,"forks":2064,"watchers":150,"issues":587,"prs":952,"releases":45,"issues_open":325,"issues_closed":262,"prs_open":61,"prs_closed":161,"prs_merged":730,"discussions":146},"growth":{"day":{"stars":255,"forks":118,"base_day":"2026-11-15"},"week":{"stars":455,"forks":155,"base_day":"2026-11-04"},"month":{"stars":548,"forks":187,"base_day":"2026-10-17"}},"history":{"dates":["2026-09-06","2026-09-07","2026-09-08","2026-09-09","2026-09-10","2026-09-11","2026-09-12","2026-09-13","2026-09-14","2026-09-15","2026-09-16","2026-09-17","2026-09-18","2026-09-19","2026-09-20","2026-09-21","2026-09-22","2026-09-23","2026-09-24","2026-09-25","2026-09-26","2026-09-27","2026-09-28","2026-09-29","2026-09-30","2026-10-01","2026-10-02","2026-10-03","2026-10-04","2026-10-05","2026-10-06","2026-10-07","2026-10-08","2026-10-09","2026-10-10","2026-10-11","2026-10-12","2026-10-13","2026-10-14","2026-10-15","2026-10-16","2026-10-17","2026-10-18","2026-10-19","2026-10-20","2026-10-21","2026-10-22","2026-10-23","2026-10-24","2026-10-25","2026-10-26","2026-10-27","2026-10-28","2026-10-29","2026-10-30","2026-10-31","2026-11-01","2026-11-02","2026-11-03","2026-11-04","2026-11-10","2026-11-11","2026-11-12","2026-11-13","2026-11-14","2026-11-15","2026-11-16"],"stars":[15295,15299,15305,15313,15319,15327,15338,15343,15346,15356,15361,15364,15367,15369,15373,15380,15384,15392,15399,15407,15413,15414,15417,15418,15423,15427,15433,15439,15440,15443,15445,15445,15451,15455,15458,15463,15471,15475,15481,15481,15484,15487,15495,15498,15505,15514,15515,15514,15520,15524,15532,15540,15544,15550,15555,15559,15564,15567,15576,15580,15601,15608,15617,15618,15624,15780,16035],"forks":[1808,1811,1813,1812,1813,1814,1815,1819,1821,1823,1825,1824,1823,1823,1823,1824,1824,1829,1831,1833,1836,1839,1840,1842,1844,1846,1849,1853,1856,1857,1858,1859,1859,1862,1866,1868,1869,1873,1876,1876,1876,1877,1877,1878,1880,1879,1882,1883,1882,1884,1885,1885,1885,1889,1894,1897,1900,1906,1907,1909,1924,1927,1929,1929,1936,1946,2064],"watchers":[153,154,154,154,154,154,154,154,154,154,154,154,154,154,155,155,155,155,155,155,155,155,155,155,155,155,155,156,156,156,156,156,156,156,156,157,157,157,157,157,156,156,156,156,156,156,156,156,156,151,151,151,151,151,150,150,150,150,150,150,null,150,150,150,150,150,150],"issues_open":[null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,326,326,326,326,326,325],"prs_open":[null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,57,58,59,60,61,61]},"boards":{"daily":92,"daily_rate":115,"daily_forks":25},"same_owner":[{"full_name":"ggml-org/llama.cpp","name":"llama.cpp","shard":"1","boards":{"daily":178,"weekly":130,"daily_forks":107,"daily_watchers":59,"total":86}},{"full_name":"ggml-org/Llama-macOS","name":"Llama-macOS","shard":"1","boards":{"daily_watchers":191,"debut":24}},{"full_name":"ggml-org/whisper.cpp","name":"whisper.cpp","shard":"1","boards":{}}],"issues":[{"number":59,"title":"Text to speech models in GGML?","comments":93,"created_at":"2023-04-01T15:10:23Z"},{"number":8,"title":"How to fine tune it?","comments":43,"created_at":"2022-12-09T10:56:01Z"},{"number":293,"title":"ggml : get rid of BLAS and all it's variants","comments":39,"created_at":"2023-06-25T09:16:12Z"},{"number":21,"title":"Suggestion: RWKV Language Model","comments":20,"created_at":"2023-02-06T04:25:14Z"},{"number":542,"title":"[Question] What is the status of Vulkan backend?","comments":17,"created_at":"2023-09-27T18:01:25Z"},{"number":436,"title":"Cannot load teknium/Replit-v2-CodeInstruct-3B","comments":14,"created_at":"2023-08-06T16:02:56Z"},{"number":6,"title":"[Feature request] Implement GPT-JT","comments":14,"created_at":"2022-12-01T10:13:12Z"},{"number":233,"title":"ggml : implement a spellcheck model (xfspell, t5-spellchecker, etc)","comments":12,"created_at":"2023-06-06T16:05:12Z"}],"discussions":[{"number":1693,"title":"Transformation from Nvidia to ggml","comments":1,"category":"General","created_at":"2026-11-14T13:51:35Z"},{"number":1686,"title":"clap.cpp for audio embedding","comments":0,"category":"Show and tell","created_at":"2026-11-09T04:16:51Z"},{"number":1674,"title":"Optimized kernels in PTX","comments":0,"category":"General","created_at":"2026-10-27T19:02:03Z"},{"number":1649,"title":"old NVIDIA GPU arch run(Fermi,Kepler) support, and LLVM-IR NVPTX(Clang) compiler support","comments":1,"category":"Show and tell","created_at":"2026-10-04T10:12:21Z"},{"number":1605,"title":"Help","comments":0,"category":"Q&A","created_at":"2026-08-24T12:06:15Z"},{"number":1579,"title":"dist : releases and versioning of `ggml-org` projects","comments":17,"category":"General","created_at":"2026-08-07T08:13:57Z"},{"number":1543,"title":"Verifying Q4_K scale decode: uint32 bit-op approach vs get_scale_min_k4 equivalence","comments":1,"category":"Q&A","created_at":"2026-06-15T06:48:11Z"},{"number":1524,"title":"use ggml in android NDK","comments":1,"category":"Q&A","created_at":"2026-05-30T17:07:29Z"}],"latest_release":{"name":"v1.0.0","tag":"v1.0.0","published_at":"2026-11-13T08:34:21Z","body":"## Overview\r\n\r\nggml 1.0.0 prunes the deprecated public API and the legacy `DIAG_MASK` ops, adds configurable YaRN truncation and FP8 support, and brings broad performance work across backends: few-row quantized matmuls and MoE fusion on CUDA/Metal, a Vulkan prefill FlashAttention path, a device-side MoE scheduler on SYCL, decomposed Adreno FlashAttention in OpenCL, new WebGPU kernels, expanded Hexagon op coverage, and scheduler improvements that avoid uploading unused inputs.\r\n\r\n### API changes\r\n\r\n- Pruned deprecated API and removed the legacy `DIAG_MASK_INF`/`DIAG_MASK_ZERO` ops, bumping the RPC protocol to reject incompatible peers ([llama/29801](https://github.com/ggml-org/llama.cpp/pull/29801), [llama/31946](https://github.com/ggml-org/llama.cpp/pull/31946))\r\n- Added `ggml_rope_set_truncate`/`ggml_rope_get_truncate` to control YaRN correction-bound rounding ([llama/30206](https://github.com/ggml-org/llama.cpp/pull/30206), [llama/30802](https://github.com/ggml-org/llama.cpp/pull/30802))\r\n- Implemented the backward pass for `TRI` ([llama/30712](https://github.com/ggml-org/llama.cpp/pull/30712))\r\n- Added `MUL_MAT` src1 precision tests and gating for `ggml_prec_set_acc` ([llama/31896](https://github.com/ggml-org/llama.cpp/pull/31896), [llama/31851](https://github.com/ggml-org/llama.cpp/pull/31851))\r\n- Bumped the RPC protocol to v9 with faster loading and graceful failure on unreachable endpoints or failed handshakes ([llama/29413](https://github.com/ggml-org/llama.cpp/pull/29413), [llama/30415](https://github.com/ggml-org/llama.cpp/pull/30415))\r\n\r\n### Core changes\r\n\r\n- Scheduler skips uploading inputs that are not read and prefetches only inputs produced by earlier splits ([llama/31419](https://github.com/ggml-org/llama.cpp/pull/31419), [llama/32100](https://github.com/ggml-org/llama.cpp/pull/32100))\r\n- MoE cache can compute a share of the cache misses on the CPU and copy the results back asynchronously ([llama/30966](https://github.com/ggml-org/llama.cpp/pull/30966))\r\n- Meta backend raises the compute view limit for buffers with few tensors ([llama/31447](https://github.com/ggml-org/llama.cpp/pull/31447))\r\n\r\n### Backend changes\r\n\r\n#### CUDA\r\n\r\n- Added few-row MMQ for batch sizes 2-24 and MMVF for BF16 few-row matmuls ([llama/31533](https://github.com/ggml-org/llama.cpp/pull/31533), [llama/31959](https://github.com/ggml-org/llama.cpp/pull/31959))\r\n- Added OCP FP8 support and fused NVFP4 gate/up matmuls with GLU in MMQ ([llama/28413](https://github.com/ggml-org/llama.cpp/pull/28413), [llama/31296](https://github.com/ggml-org/llama.cpp/pull/31296))\r\n- Extended FlashAttention with an sm70 head-size-256 config and MMA head sizes 40/72 ([llama/27997](https://github.com/ggml-org/llama.cpp/pull/27997), [llama/30285](https://github.com/ggml-org/llama.cpp/pull/30285))\r\n- Improved graph handling: skip compat checks on reuse, re-instantiate on cluster-dim changes, move graph state ([llama/32004](https://github.com/ggml-org/llama.cpp/pull/32004), [llama/31284](https://github.com/ggml-org/llama.cpp/pull/31284), [llama/32001](https://github.com/ggml-org/llama.cpp/pull/32001))\r\n- MoE: skip empty MMQ stream-k tiles, linearize the MoE layout, and sanitize post-bias NaNs ([llama/32067](https://github.com/ggml-org/llama.cpp/pull/32067), [llama/31047](https://github.com/ggml-org/llama.cpp/pull/31047), [llama/31005](https://github.com/ggml-org/llama.cpp/pull/31005))\r\n- Added a debug buffer pool and halved the VMM pool reservation with reserve retries ([llama/31919](https://github.com/ggml-org/llama.cpp/pull/31919), [llama/30542](https://github.com/ggml-org/llama.cpp/pull/30542))\r\n- HIP: use a fixed BLAS workspace and add an MFMA lightning indexer for CDNA2 ([llama/31838](https://github.com/ggml-org/llama.cpp/pull/31838), [llama/29050](https://github.com/ggml-org/llama.cpp/pull/29050))\r\n- Added a d_state=64 SSM_SCAN fast path and made radix Top-K deterministic for ties ([llama/31818](https://github.com/ggml-org/llama.cpp/pull/31818), [llama/31084](","is_prerelease":false},"readme":true,"readme_zh":false,"news":[],"stale":false,"snapshot_at":"2026-11-16T09:23:14Z","history_start":"2026-09-06"}