bukejiyu 598cce8545 [RL] Support SM100 FP8 quantization in RL (#6601) ...

* RL SM100 Fix

* update

2026-03-04 04:55:04 -08:00

..

__init__.py

support w4afp8 EP inference (#3044)

2025-08-25 11:27:45 +08:00

ep.py

fix pfcc deep ep in low latency mode (#6440)

2026-03-02 10:35:51 +08:00

fused_moe_backend_base.py

[Feature] Support redundant expert for eplb (#5918)

2026-01-09 17:13:24 +08:00

fused_moe_cutlass_backend.py

[Iluvatar] Support CudaGraph and optimize flash_attn_unpadded and fused_neox_rope_embedding (#6553)

2026-03-02 14:07:17 +08:00

fused_moe_deepgemm_backend.py

[Quantization] Support to load static quant ue8m0 scale of DeepGEMM via v0_loader (#6433)

2026-03-03 11:32:35 +08:00

fused_moe_marlin_backend.py

[Optimization] Enable BF16 gate computation for GLM and Qwen (#6457)

2026-02-26 21:08:46 -08:00

fused_moe_triton_backend.py

[RL] Support SM100 FP8 quantization in RL (#6601)

2026-03-04 04:55:04 -08:00

fused_moe_wint2_backend.py

[loader]supoort wint2 backend (#6139)

2026-02-08 22:42:36 -08:00

moe.py

[loader]supoort wint2 backend (#6139)

2026-02-08 22:42:36 -08:00

routing_indices_cache.py

[RL] R3 Support Fused Put the Routing of All Layers (#6099)

2026-02-03 04:13:16 -08:00

triton_moe_kernels.py

[OPs] MoE support wfp8afp8(channelwise) and improve per_token_quant_fp8 (#4238)

2025-09-24 16:39:51 +08:00