Артем Савкин b77b0ffd60 [NPU] NZ for non-quantized MOE, Qwen3 MOE double memory consumption fix (#15904) ...

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

2026-01-29 00:55:08 +08:00

..

attention

[DSv32] Overlap indexer qk projection and activation quant (#17688)

2026-01-28 11:46:49 +08:00

deep_gemm_wrapper

[fix]deepgemm precompile when warmup (#15891)

2025-12-28 13:47:39 +08:00

moe

[NPU] NZ for non-quantized MOE, Qwen3 MOE double memory consumption fix (#15904)

2026-01-29 00:55:08 +08:00

quantization

[NPU] NZ for non-quantized MOE, Qwen3 MOE double memory consumption fix (#15904)

2026-01-29 00:55:08 +08:00

utils

[Refactor] Rename CustomOp -> MultiPlatformOp (#16175)

2025-12-31 01:16:32 +08:00

activation.py

[cpu/arm64] support run sglang on arm64 cpu (#14867)

2026-01-11 04:27:19 -08:00

amx_utils.py

[CPU] Optimize small oc GEMM for Qwen3-next on CPU (#12446)

2025-12-04 00:38:47 -08:00

communicator_nsa_cp.py

[DeepSeek v3.2] opt Context Parallelism: support fused moe, multi batch and fp8 kvcache (#13959)

2026-01-02 23:49:14 +08:00

communicator.py

[hotfix] Reenable all reduce fusion on sm100 (#17591)

2026-01-22 23:36:38 +08:00

dp_attention.py

Use attn_tp_group for all reduce in token embedding (#17403)

2026-01-20 23:56:21 +08:00

elementwise.py

[Refactor] Clean up custom op (#15995)

2026-01-01 21:41:56 +08:00

flashinfer_comm_fusion.py

[Refactor] Clean up custom op (#15995)

2026-01-01 21:41:56 +08:00

int4fp8_utils.py

[AMD][Quantization] Add int4fp8_moe online quantization on ROCm (#7392)

2026-01-14 01:44:40 -08:00

layernorm.py

fix post_residual_addition more generally (#17286)

2026-01-23 15:43:37 -08:00

linear.py

Fix circular import in quantization modules (#17372)

2026-01-21 15:47:09 +08:00

logits_processor.py

[Auto Sync] Update logits_processor.py, test_logprobs.py (20260124) (#17664)

2026-01-23 17:57:41 -08:00

model_parallel.py

[Refactor] Rename CustomOp -> MultiPlatformOp (#16175)

2025-12-31 01:16:32 +08:00

modelopt_utils.py

Enable native ModelOpt quantization support (1/3) (#7149)

2025-10-06 13:24:15 -07:00

multimodal.py

[VLM] Improving multimodal tensor hash kernel (#9008)

2025-08-14 13:45:55 -07:00

parameter.py

Code clean up for fp8 quantization (#16982)

2026-01-13 12:38:39 -08:00

pooler.py

Support piecewise CUDA graph for embedding models (#13852)

2025-11-26 15:29:46 +08:00

radix_attention.py

tiny refactor pcg split op registration (#16863)

2026-01-10 23:45:28 +08:00

radix_linear_attention.py

[hybrid-model] clean up and consolidate redundant fields in RadixLinearAttention (#17660)

2026-01-27 10:37:58 -08:00

rocm_linear_utils.py

Aiter fp8 kv cache (#13147)

2025-12-08 16:39:53 -08:00

rotary_embedding.py

[Model] Add Ernie4.5 VL model support (#15679)

2026-01-25 22:36:29 -08:00

sampler.py

Revert "feat: reduce constrained-decoding overhead in TP" (#16845)

2026-01-10 11:39:38 +08:00

sparse_pooler.py

Implement BGE-M3 Sparse Embeddings in SGLang (#10869)

2025-10-22 13:46:16 -07:00

torchao_utils.py

Lazy import torchao (#17626)

2026-01-22 22:04:51 -08:00

vocab_parallel_embedding.py

Use attn_tp_group for all reduce in token embedding (#17403)

2026-01-20 23:56:21 +08:00