gnovackandJee Jee Li 294c805f1d Early exit for MoE LoRA kernels (#27131) ...

Signed-off-by: gnovack <gnovack@amazon.com>
Co-authored-by: Jee Jee Li <pandaleefree@gmail.com>

2025-11-03 20:22:17 +08:00

..

attention

[Attention] Tune CUTLASS MLA num_splits (#26846)

2025-10-16 06:36:09 -07:00

core

[small][batch invariance] Rename the env and internal flags to simplify usage (#26855)

2025-10-16 21:40:25 +00:00

cpu

[cpu][fix] Fix onednn_mm crash on consecutive matmuls with same M,K,N and different dtype (#27472)

2025-10-24 15:57:48 +00:00

cutlass_extensions

Update Optional[x] -> x | None and Union[x, y] to x | y (#26633)

2025-10-12 09:51:31 -07:00

mamba/mamba_ssm

[V1] [Hybrid] Mamba1 Automatic Prefix Caching (#26377)

2025-11-02 04:16:23 -08:00

moe

Early exit for MoE LoRA kernels (#27131)

2025-11-03 20:22:17 +08:00

quantization

[Kernel] Add GPTQv2 format support for low-bit or asymmetric quantization, by adapting gptq_gemm (#26092)

2025-10-23 23:26:13 -04:00

quickreduce

[Bugfix][Rocm] fix qr error when different inp shape (#25892)

2025-10-13 10:04:21 -07:00

rocm

[Refactor] Refactor FP8 & INT8 Quant Folder inside w8a8 (#25293)

2025-10-08 10:20:48 -04:00

sparse/cutlass

…

activation_kernels.cu

…

cache_kernels.cu

[Refactor] Refactor FP8 & INT8 Quant Folder inside w8a8 (#25293)

2025-10-08 10:20:48 -04:00

cache.h

Add gather_indexer_k_quant_cache kernel (#25931)

2025-10-08 04:58:57 +00:00

cub_helpers.h

[Refactor] Refactor FP8 & INT8 Quant Folder inside w8a8 (#25293)

2025-10-08 10:20:48 -04:00

cuda_compat.h

…

cuda_utils_kernels.cu

…

cuda_utils.h

…

cuda_view.cu

…

cumem_allocator.cpp

…

custom_all_reduce_test.cu

…

custom_all_reduce.cu

…

custom_all_reduce.cuh

…

custom_quickreduce.cu

…

dispatch_utils.h

…

launch_bounds_utils.h

Update launch_bounds_utils.h for correct compile on Multiple Cuda Arch - PTXAS out of range Warning (#25843)

2025-09-30 19:18:19 -07:00

layernorm_kernels.cu

[Chore] Remove unused PolyNorm layer (#27110)

2025-10-17 19:03:43 +00:00

layernorm_quant_kernels.cu

[torch.compile] Enable attention and allreduce fusion without custom ops enabled (#24604)

2025-10-17 08:10:23 -06:00

ops.h

[V1] [Hybrid] Mamba1 Automatic Prefix Caching (#26377)

2025-11-02 04:16:23 -08:00

permute_cols.cu

…

pos_encoding_kernels.cu

[Chore] Remove unused batched RoPE op & kernel (#24789)

2025-09-13 00:08:20 -07:00

sampler.cu

[Deepseek v3.2] Remove extra logics in indexer (#26465)

2025-10-21 23:34:03 +00:00

torch_bindings.cpp

[V1] [Hybrid] Mamba1 Automatic Prefix Caching (#26377)

2025-11-02 04:16:23 -08:00

type_convert.cuh

…