Lucas Wilkinson 7eb4255628

[BugFix] Accuracy fix for llama4 int4 - improperly casted scales (#16801) ...

Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>

2025-04-17 22:13:29 -07:00

..

attention

[Bugfix][Kernel] fix potential cuda graph broken for merge_attn_states kernel (#16693)

2025-04-16 03:31:39 -07:00

core

[Attention] MLA with chunked prefill (#12639)

2025-02-21 15:30:12 -08:00

cpu

[Bugfix] fix gettid method is not define (#16084)

2025-04-08 19:12:44 -07:00

cutlass_extensions

[Kernel] CUTLASS grouped gemm fp8 MoE kernel (#13972)

2025-03-27 00:54:44 +00:00

mamba

[BugFix] fix some typos found by typos. (#16314)

2025-04-09 03:43:59 -07:00

moe

[BugFix] Accuracy fix for llama4 int4 - improperly casted scales (#16801)

2025-04-17 22:13:29 -07:00

prepare_inputs

[Misc][Easy] Annotate unused vars in the csrc files (#14798)

2025-03-15 12:40:09 +08:00

quantization

[Kernel] moe wna16 marlin kernel (#14447)

2025-04-14 20:05:22 -07:00

rocm

[ROCM][KERNEL] Paged attention for V1 (#15720)

2025-04-02 19:48:00 -07:00

sparse/cutlass

[BugFix/Build] Fix sparse kernels not getting built on hopper (#14572)

2025-03-11 17:09:03 +00:00

activation_kernels.cu

[Kernel] Support MulAndSilu (#11624)

2025-01-15 02:29:53 +00:00

cache_kernels.cu

[Misc][Docs] fix the comments of KV_T and CACHE_T in CALL_RESHAPE_AND_CACHE_XX macros (#14347)

2025-03-18 05:50:19 -07:00

cache.h

[Attention] MLA with chunked prefill (#12639)

2025-02-21 15:30:12 -08:00

cuda_compat.h

[Kernel][ROCm][AMD] enable fused topk_softmax kernel for moe layer (#4927)

2024-06-02 14:13:26 -07:00

cuda_utils_kernels.cu

[NVIDIA] Support nvfp4 quantization (#12784)

2025-02-12 19:51:51 -08:00

cuda_utils.h

[Attention] MLA with chunked prefill (#12639)

2025-02-21 15:30:12 -08:00

cuda_view.cu

[V1] Fully Transparent Implementation of CPU Offloading (#15354)

2025-03-31 20:22:34 +08:00

cumem_allocator.cpp

[core] improve error handling when wake up from sleep mode (#12981)

2025-02-10 09:38:57 +08:00

custom_all_reduce_test.cu

[Distributed] Add custom allreduce support for ROCM (#14125)

2025-03-31 22:49:12 -07:00

custom_all_reduce.cu

[Distributed] Add custom allreduce support for ROCM (#14125)

2025-03-31 22:49:12 -07:00

custom_all_reduce.cuh

fix: spelling (#16466)

2025-04-11 23:24:22 -07:00

dispatch_utils.h

dynamic distpatch of fp8 kernels (#14245)

2025-03-11 10:54:56 -04:00

layernorm_kernels.cu

[torch.compile] Fuse RMSNorm with quant (#9138)

2024-11-08 21:20:08 +00:00

layernorm_quant_kernels.cu

dynamic distpatch of fp8 kernels (#14245)

2025-03-11 10:54:56 -04:00

ops.h

[Kernel] support merge_attn_states CUDA kernel, 3x speedup (#16173)

2025-04-11 06:50:50 -06:00

permute_cols.cu

[Kernel] (2/N) Machete - Integrate into CompressedTensorsWNA16 and GPTQMarlin (#7701)

2024-09-23 13:46:26 -04:00

pos_encoding_kernels.cu

[Kernel] Make rotary_embedding ops more flexible with input shape (#12777)

2025-02-06 08:46:13 -08:00

torch_bindings.cpp

[Kernel] support merge_attn_states CUDA kernel, 3x speedup (#16173)

2025-04-11 06:50:50 -06:00

type_convert.cuh

[torch.compile] Fuse RMSNorm with quant (#9138)

2024-11-08 21:20:08 +00:00