Default Branch

master c7fc509417 · opencl: add F32-to-I32 ARGMAX (#30897) · Updated 2026-10-22 09:02:50 +08:00

ZIP  TAR.GZ

Branches

gg/llama-high-throughput-save2 6179578988 · batch : require non-coupled batch with sequential split_equal · Updated 2025-06-25 22:20:46 +08:00    tqcq	5463 29		ZIP  TAR.GZ
gg/llama-high-throughput-save 37bdfbef8c · wip 3 · Updated 2025-06-24 16:04:18 +08:00    tqcq	5463 21		ZIP  TAR.GZ
gg/metal-fix-thread-safety ae96333923 · metal : fix thread-safety · Updated 2025-06-20 21:42:54 +08:00    tqcq	5489 1		ZIP  TAR.GZ
gg/repack-fix-wsize 6fb2f2e8a9 · ggml : fix repack work size for mul_mat_id · Updated 2025-06-20 15:34:16 +08:00    tqcq	5492 1		ZIP  TAR.GZ
gabe-l-hart/HybridRecurrentCache 59fee24c72 · recurrent : rework graph inputs + add TODOs · Updated 2025-06-18 14:29:51 +08:00    tqcq	5516 31		ZIP  TAR.GZ
add-pidfile d3d06debe3 · server : add pidfile option · Updated 2025-06-18 04:47:53 +08:00    tqcq	5517 1		ZIP  TAR.GZ
gg/server-reenable-swa-spec 36fce98281 · server : re-enable swa speculative decoding · Updated 2025-06-12 16:51:15 +08:00    tqcq	5560 1		ZIP  TAR.GZ
gg/batch-simplify-output ed99a8ea04 · cont : fix comments · Updated 2025-06-12 15:43:55 +08:00    tqcq	5563 3		ZIP  TAR.GZ
compilade/fix-batch-reserve-rwkv 4b6fb6524b · context : round n_tokens to next multiple of n_seqs when reserving · Updated 2025-06-12 04:19:17 +08:00    tqcq	5566 1		ZIP  TAR.GZ
compilade/readonly-recurrent-inputs 62a9f34bae · llama-graph : fix recurrent state copy · Updated 2025-06-10 12:26:30 +08:00    tqcq	5605 3		ZIP  TAR.GZ
gg/kv-fix-shift c257a8871c · cont : fix defrag erasing cells that didn't move · Updated 2025-06-10 01:45:56 +08:00    tqcq	5612 3		ZIP  TAR.GZ
maxk/sched-prio-updates ac35e50c16 · Update tools/llama-bench/llama-bench.cpp · Updated 2025-06-01 06:38:37 +08:00    tqcq	5672 3		ZIP  TAR.GZ
cisc/test-tokenizers-remote d3a2eb592d · disable on windows · Updated 2025-06-01 05:17:18 +08:00    tqcq	5663 12		ZIP  TAR.GZ
gg/min-p-fix 9065ca71a2 · tests : sampling tests use min_keep == 0 · Updated 2025-05-27 16:30:41 +08:00    tqcq	5718 3		ZIP  TAR.GZ
gg/tts-fix-ubatch 108d484ab2 · tts : fix n_ubatch + make WavTokenizer cache-less · Updated 2025-05-23 02:58:10 +08:00    tqcq	5762 1		ZIP  TAR.GZ
jared/permit-causal-encode b06a954bbc · llama_encode : only force non-causal attention for enc-dec models · Updated 2025-05-20 01:43:59 +08:00    tqcq	5795 1		ZIP  TAR.GZ
gg/bench-handle-decode-errors 8282d74692 · bench : handle decode errors · Updated 2025-05-15 03:36:29 +08:00    tqcq	5833 1		ZIP  TAR.GZ
gg/server-models-loading 237acc7cd5 · server : update readme + return json for "meta" field · Updated 2025-05-14 20:30:12 +08:00    tqcq	5842 2		ZIP  TAR.GZ
gg/metal-fa-vec-bs20 78d70223c3 · metal : use FA-vec kernel up to batch size 20 · Updated 2025-05-13 15:38:06 +08:00    tqcq	5859 3		ZIP  TAR.GZ
gg/context-remove-logits-all 6107303ab0 · llama : remove logits_all flag + reorder llama_context_params · Updated 2025-05-08 18:01:41 +08:00    tqcq	5912 2		ZIP  TAR.GZ