分支列表

默认分支

main
pre-commit / pre-run-check (push) Has been cancelled
pre-commit / pre-commit (push) Has been cancelled

46e6a83ce1 · [Bugfix][KV Offload] Clean up resources after initialization failure (#51227) · 最后更新于 2026-08-06 11:42:13 +00:00

分支列表

e329dbcb0f · [Build] Update DeepEP pin for low-latency fabric support · 最后更新于 2026-04-19 01:06:19 +00:00    frostbyte_neo

3729
3

e329dbcb0f · [Build] Update DeepEP pin for low-latency fabric support · 最后更新于 2026-04-19 01:06:19 +00:00    frostbyte_neo

3729
3

b1388b1fbf · [CI][NIXL] Fix PD CI breakage: pin nixl-cu{12,13} versions (#39851) · 最后更新于 2026-04-18 00:57:52 +00:00    frostbyte_neo

4280
32

a60418e6fb · Sparse MLA on Hopper: Use SGLang's kernel for the sparse mla low latency runs · 最后更新于 2026-04-17 16:51:32 +00:00    frostbyte_neo

3754
1

a93a758a6c · Sparse MLA on Hopper: Use SGLang's kernel for the sparse mla low latency runs · 最后更新于 2026-04-17 16:29:06 +00:00    frostbyte_neo

3754
1

f3550b8410 · fuse more · 最后更新于 2026-04-17 08:05:36 +00:00    frostbyte_neo

3766
67

cefb933410 · perf(deepseek): Multi-stream overlap of indexer wk+weights_proj with QKV-A for DeepSeek V3.2 NSA · 最后更新于 2026-04-16 21:50:16 +00:00    frostbyte_neo

3780
1

5d236e5cf8 · [Bugfix] Add MigCudaPlatform to avoid CUDA init before fork · 最后更新于 2026-04-16 18:49:08 +00:00    frostbyte_neo

3832
4

aa0db604c1 · fix benchmark · 最后更新于 2026-04-15 21:24:53 +00:00    frostbyte_neo

3875
67

3beb57a238 · [XPU] properly handle q_descale on XPU as quant query input not supported (#39676) · 最后更新于 2026-04-15 13:52:58 +00:00    frostbyte_neo

3822
0
已包含

0442d1e57f · [CI] Reorganize release pipeline: separate nightly vs release sections · 最后更新于 2026-04-15 08:46:00 +00:00    frostbyte_neo

3831
3

f478d42cdb · [CI] Reorganize release pipeline: separate nightly vs release sections · 最后更新于 2026-04-15 08:08:39 +00:00    frostbyte_neo

3831
2

71b081ff19 · Use instanttensor == 0.1.8 · 最后更新于 2026-04-14 17:11:57 +00:00    frostbyte_neo

3881
9

8bf46ccf97 · tweaks · 最后更新于 2026-04-13 22:03:17 +00:00    frostbyte_neo

3907
2

6d4a8e6d20 · p · 最后更新于 2026-04-10 03:21:59 +00:00    frostbyte_neo

4280
19

4c3e22ad20 · update · 最后更新于 2026-04-09 16:57:23 +00:00    frostbyte_neo

3983
3

9b134c72d3 · fix v2 is_prefilling · 最后更新于 2026-04-08 19:24:06 +00:00    frostbyte_neo

4024
1

bcb5e98f11 · [Dependencies] Upgrade transformers to >= 5.5.0 and compressed-tensors to >= 0.15.0 · 最后更新于 2026-04-08 18:30:50 +00:00    frostbyte_neo

4017
1

43a9b1afbb · p · 最后更新于 2026-04-07 09:49:14 +00:00    frostbyte_neo

4280
14

c5e3454e5a · [Model] Add support for BharatGen's Param2MoE model (#38000) · 最后更新于 2026-04-06 08:19:56 +00:00    frostbyte_neo

4082
0
已包含