Default Branch

master 1c4729414d · hex-scripts: show trace events smaller than 100nsec in perfetto (#29614) · Updated 2026-09-28 17:02:58 -05:00

ZIP  TAR.GZ

Branches

ceb/fix-logit-check 1b05817112 · decode : fix logits_valid for old API · Updated 2023-12-17 17:49:21 -06:00    upstream-archive	9586 1		ZIP  TAR.GZ
gg/swiftui-bench 865066621b · llama.swiftui : improve bench · Updated 2023-12-17 11:37:22 -06:00    upstream-archive	9600 12		ZIP  TAR.GZ
pr/4484 f86b9d152c · lookup : minor · Updated 2023-12-17 09:25:28 -06:00    upstream-archive	9598 9		ZIP  TAR.GZ
gg/phi-2 d2f1e0dacc · Merge branch 'cuda-cublas-opts' into gg/phi-2 · Updated 2023-12-17 00:41:46 -06:00    upstream-archive	9596 17		ZIP  TAR.GZ
ceb/fix-badspecial-silentfail b0547d2196 · gguf-py : fail fast on nonsensical special token IDs · Updated 2023-12-15 17:06:42 -06:00    upstream-archive	9598 1		ZIP  TAR.GZ
ceb/fix-cuda-warning-flags c8554b80be · Merge branch 'master' of https://github.com/ggerganov/llama.cpp into ceb/fix-cuda-warning-flags · Updated 2023-12-13 11:06:01 -06:00    upstream-archive	9610 12		ZIP  TAR.GZ
mixtral e1241d9b46 · metal : switch to execution barriers + fix one of the barriers · Updated 2023-12-13 05:56:45 -06:00    upstream-archive	9621 47		ZIP  TAR.GZ
gg/per-layer-kv fc5f334689 · readme : add API change notice · Updated 2023-12-07 04:35:02 -06:00    upstream-archive	9623 15		ZIP  TAR.GZ
gg/quantum-k-cache af99c6fbfc · llama : remove memory_f16 and kv_f16 flags · Updated 2023-12-05 10:18:16 -06:00    upstream-archive	9635 26		ZIP  TAR.GZ
gg/pad-kv-cache 3cb1c348b3 · metal : try to improve batched decoding · Updated 2023-12-01 14:01:58 -06:00    upstream-archive	9640 2		ZIP  TAR.GZ
gg/soft-max-ext eb594c0f7d · alloc : fix build with debug · Updated 2023-12-01 02:46:05 -06:00    upstream-archive	9664 14		ZIP  TAR.GZ
ceb/libstdcpp-assertions 5b74310e6e · build : enable libstdc++ assertions for debug builds · Updated 2023-11-30 17:18:24 -06:00    upstream-archive	9649 1		ZIP  TAR.GZ
assert-restore-abort bb39b87964 · ggml : restore abort() in GGML_ASSERT · Updated 2023-11-27 18:27:09 -06:00    upstream-archive	9668 1		ZIP  TAR.GZ
gg/fix-cpu-blas 87f4102a70 · llama : revert n_threads_batch logic · Updated 2023-11-27 13:47:35 -06:00    upstream-archive	9669 3		ZIP  TAR.GZ
ceb/perf-faster-multigpu 6272b6764a · use stride=128 if built for tensor cores · Updated 2023-11-27 12:09:14 -06:00    upstream-archive	9672 3		ZIP  TAR.GZ
lookahead 8d8b76d469 · lookahead : add comments · Updated 2023-11-26 03:26:55 -06:00    upstream-archive	9684 9		ZIP  TAR.GZ
server-oai-compat 21b70babf7 · straightforward /v1/models endpoint · Updated 2023-11-24 10:22:39 -06:00    upstream-archive	9685 12		ZIP  TAR.GZ
kv-cache-opts f8e9f11428 · common : add -dkvc arg for enabling kv cache dumps · Updated 2023-11-23 10:47:56 -06:00    upstream-archive	9691 4		ZIP  TAR.GZ
ceb/fix-yarn-neox f824902623 · YaRN : correction to GPT-NeoX implementation · Updated 2023-11-15 16:10:52 -06:00    upstream-archive	9723 1		ZIP  TAR.GZ
llama-metadata d0445a2eff · better documentation · Updated 2023-11-09 18:38:20 -06:00    upstream-archive	9740 3		ZIP  TAR.GZ