Daniel Han b8eee7a8ba Revert "Initial changes: Refactor Attention" ... This reverts commit a2af843271.		2026-03-13 22:38:57 -07:00
..
moe	fix(ROCm): prevent false TMA support detection on AMD GPUs (#4126)	2026-02-28 23:59:27 -08:00
__init__.py	Formatting & bug fixes (#3563)	2025-11-07 06:00:22 -08:00
cross_entropy_loss.py	Revert "perf(ROCm): optimize chunked CE loss num_warps for RDNA GPUs (#4123)" (#4139)	2026-03-03 03:05:32 -08:00
fast_lora.py	[pre-commit.ci] auto fixes from pre-commit.com hooks	2025-12-30 15:58:41 +00:00
flex_attention.py	Revert "Initial changes: Refactor Attention"	2026-03-13 22:38:57 -07:00
fp8.py	Suppress FBGEMM CUTLASS stdout spam on Blackwell GPUs (#4092)	2026-02-23 01:27:10 -08:00
geglu.py	Revert "[FIX] Vllm guided decoding params (#3662)"	2025-12-01 05:43:45 -08:00
layernorm.py	fix(Triton): ensure float32 eps in RMS LayerNorm rsqrt for HIP/ROCm (#4110)	2026-02-28 23:59:22 -08:00
rms_layernorm.py	fix(Triton): ensure float32 eps in RMS LayerNorm rsqrt for HIP/ROCm (#4110)	2026-02-28 23:59:22 -08:00
rope_embedding.py	fastrope fix for zero strided tensors (#3782)	2025-12-28 21:21:48 -08:00
swiglu.py	Apply suggestion from @danielhanchen	2026-01-12 21:32:20 -08:00
utils.py	fix(ROCm): restrict is_rdna() to ROCm-officially-supported RDNA GPUs (#4136)	2026-03-03 03:05:38 -08:00