Default Branch

master d775ebf363 · server: return HTTP 400 for invalid embedding requests (#29060) · Updated 2026-10-01 09:52:43 -05:00

ZIP  TAR.GZ

Branches

fix_q_xxs_mul_mat 4b1770109c · Fix q_xxs using mul_mat_q · Updated 2024-05-27 03:46:37 -05:00    upstream-archive	8304 1		ZIP  TAR.GZ
gg/metal-disable-fa-256 1c6cde92bb · metal : disable FA kernel for HS=256 · Updated 2024-05-27 01:57:20 -05:00    upstream-archive	8306 1		ZIP  TAR.GZ
compilade/lazier-moe-convert-hf 11f78c6a2d · convert-hf : adapt ArcticModel to use yield too · Updated 2024-05-25 11:52:53 -05:00    upstream-archive	8313 4		ZIP  TAR.GZ
7507-main-intel-dockerfile dd14d818e0 · Update main-intel.Dockerfile base image to 2024.1.0 · Updated 2024-05-23 21:47:58 -05:00    upstream-archive	8324 1		ZIP  TAR.GZ
compilade/gguf-py-fix-q-shape c5fe1d6cdc · gguf-py : remove unused import · Updated 2024-05-22 23:09:49 -05:00    upstream-archive	8339 2		ZIP  TAR.GZ
sl/cuda-uma 518b75260b · cuda uma test · Updated 2024-05-22 20:13:48 -05:00    upstream-archive	8339 1		ZIP  TAR.GZ
sl/dio-test e9095e6098 · async direct io per tensor test · Updated 2024-05-21 18:08:52 -05:00    upstream-archive	8358 3		ZIP  TAR.GZ
gg/kv-determinism a041ced0fd · wip · Updated 2024-05-20 10:20:49 -05:00    upstream-archive	8364 1		ZIP  TAR.GZ
ci-android 007f2ece0a · cmake : provide binary dir · Updated 2024-05-18 02:50:46 -05:00    upstream-archive	8399 4		ZIP  TAR.GZ
gg/test-embd 6b2f496409 · wip · Updated 2024-05-17 06:01:09 -05:00    upstream-archive	8407 1		ZIP  TAR.GZ
gg/test-bench a085a8323a · tmp · Updated 2024-05-16 06:45:47 -05:00    upstream-archive	8409 1		ZIP  TAR.GZ
sl/async-weight-copy 5de9b743f8 · sched : support async weight copy · Updated 2024-05-15 17:47:40 -05:00    upstream-archive	8416 1		ZIP  TAR.GZ
fix-convert-modelname 284870c868 · Merge branch 'master' into fix-convert-modelname · Updated 2024-05-14 00:05:49 -05:00    upstream-archive	8439 2		ZIP  TAR.GZ
sl/disable-pp-nkvo 94061d58e7 · llama : disable pipeline parallelism with nkvo · Updated 2024-05-13 16:17:53 -05:00    upstream-archive	8439 1		ZIP  TAR.GZ
mlx-challenge 33a004e9cc · llama : more metal-friendly KV cache PAD · Updated 2024-05-13 02:32:19 -05:00    upstream-archive	8445 1		ZIP  TAR.GZ
compilade/q8_0-convert-hf 65a1a58562 · convert-hf : add missing ftype to Baichuan and Xverse · Updated 2024-05-12 11:56:03 -05:00    upstream-archive	8458 3		ZIP  TAR.GZ
gg/refactor-alibi-2 03e940cdec · convert : fix convert for refact models · Updated 2024-05-11 02:31:52 -05:00    upstream-archive	8474 10		ZIP  TAR.GZ
compilade/lazy-bfloat16-convert-hf e0af2df690 · convert-hf : support outtype templating in outfile name · Updated 2024-05-10 12:42:03 -05:00    upstream-archive	8487 7		ZIP  TAR.GZ
gg/metal-fattn-reqs fecb81e302 · metal : fix ggml_metal_supports_op · Updated 2024-05-09 06:01:03 -05:00    upstream-archive	8486 2		ZIP  TAR.GZ
ceb/fix-cmake-typo 494f70f939 · cmake : fix typo · Updated 2024-05-08 15:24:02 -05:00    upstream-archive	8435 1		ZIP  TAR.GZ