Default Branch

main b3f50180ec · Read half-aligned Qwen Q2 weight rows correctly on Metal · Updated 2026-11-03 23:55:02 +08:00

ZIP  TAR.GZ

Branches

glm-5.3-flash b1b4ea0364 · cuda: accelerate GLM prefill on DGX Spark · Updated 2026-08-31 05:47:28 +08:00    mtfyun	533 0	Included	ZIP  TAR.GZ
ds4f-mxfp4 80df56af40 · Refresh DeepSeek V4 Flash 0731 quality fixtures · Updated 2026-08-01 17:46:40 +08:00    mtfyun	671 0	Included	ZIP  TAR.GZ
laguna-s2.1 448d5695d1 · Tune Laguna sampling defaults · Updated 2026-07-27 18:56:54 +08:00    mtfyun	727 17	Compare	ZIP  TAR.GZ
glm5.2 bd89932c4a · Merge remote-tracking branch 'macm5us/deepseek4-pro-correctness' into glm5.2 · Updated 2026-07-03 15:40:42 +08:00    mtfyun	741 9	Compare	ZIP  TAR.GZ
responses-api acb40bf597 · Fix Responses API live continuation · Updated 2026-05-14 07:54:00 +08:00    mtfyun	984 0	Included	ZIP  TAR.GZ