Default Branch

master 07a0c4ba92 · Revert "ggml-cuda: use CMAKE_CUDA_ARCHITECTURES if set when GGML_NATIVE=ON (#18413)" (#18426) · Updated 2025-12-28 04:53:36 -08:00

ZIP  TAR.GZ

Branches

Search

gg/float-pos 608f449880 · swift : fix build · Updated 2024-02-23 09:02:09 -08:00    happyz	5313 4		ZIP  TAR.GZ
gg/py-minor-fixes 56c047156a · py : minor fixes · Updated 2024-02-22 09:22:56 -08:00    happyz	5322 1		ZIP  TAR.GZ
sl/fix-quant-kv-shift 5271c75666 · llama : fix K-shift with quantized K (wip) · Updated 2024-02-21 15:28:42 -08:00    happyz	5330 1		ZIP  TAR.GZ
gg/flash-attn-sync f249c997a8 · llama : adapt to F16 KQ_pos · Updated 2024-02-19 03:31:02 -08:00    happyz	5368 62		ZIP  TAR.GZ
gg/metal-batched 412735ec70 · Merge branch 'master' into gg/metal-batched · Updated 2024-02-19 01:25:24 -08:00    happyz	5368 6		ZIP  TAR.GZ
gg/rename-n_ctx 47c662b0de · fix some spaces added by IDE in math op · Updated 2024-02-18 12:40:35 -08:00    happyz	5378 4		ZIP  TAR.GZ
gg/fix-android 974e3cadff · ggml : try another fix · Updated 2024-02-17 08:14:35 -08:00    happyz	5397 2		ZIP  TAR.GZ
gg/hf e856bfed3b · hf : add support for --repo and --file · Updated 2024-02-15 05:05:15 -08:00    happyz	5411 3		ZIP  TAR.GZ
ceb/nomic-bert ccd757a174 · convert : fix mistakes from refactoring · Updated 2024-02-13 09:01:30 -08:00    happyz	5419 4		ZIP  TAR.GZ
ik/iq1_s 5c977221d2 · iq1_s: slightly faster dot product · Updated 2024-02-13 05:18:27 -08:00    happyz	5425 15		ZIP  TAR.GZ
ik/fix_warnings 4246b71ad7 · Fix compiler warnings (shadow variable) · Updated 2024-02-12 22:44:56 -08:00    happyz	5428 1		ZIP  TAR.GZ
ceb/bert 7286b83d3f · BERT WIP · Updated 2024-02-06 14:10:11 -08:00    happyz	5481 1		ZIP  TAR.GZ
gg/convert-fix-byte-tokens adcf16fd68 · py : fix empty bytes arg · Updated 2024-02-05 09:53:07 -08:00    happyz	5491 2		ZIP  TAR.GZ
ik/ggml-quants-cpp 91c453fb11 · One cannot possibly be defining static_assert in a C++ compilation · Updated 2024-02-05 03:22:14 -08:00    happyz	5496 2		ZIP  TAR.GZ
gg/flash-attn-interleave-cc 49a483e0f2 · wip · Updated 2024-02-04 02:34:36 -08:00    happyz	5522 60		ZIP  TAR.GZ
gg/flash-attn-32x8 a647257b47 · cuda : express strides with helper constants · Updated 2024-02-04 01:45:26 -08:00    happyz	5522 60		ZIP  TAR.GZ
gg/flash-attn-cuda b957b8f5f6 · cuda : add flash_attn kernel (wip) · Updated 2024-02-01 09:49:57 -08:00    happyz	5526 39		ZIP  TAR.GZ
flash-attn-cuda ac26f27028 · cuda : increase C to 128 for better performance · Updated 2024-02-01 07:08:29 -08:00    happyz	5526 61		ZIP  TAR.GZ
gg/flash-attn-mask-f16 1ad42b1f1e · ggml : ggml_soft_max uses F16 mask · Updated 2024-01-31 10:33:59 -08:00    happyz	5526 36		ZIP  TAR.GZ
ik/fix_iq3xxs_metal 719a087138 · iq3_xxs: forgotten update of the grid points · Updated 2024-01-30 08:39:07 -08:00    happyz	5540 1		ZIP  TAR.GZ