akhoroshev 4585f8eb95 [refactor] remove unused op_mlp (#26673)

2026-05-29 02:38:56 -07:00

..

deepseek_common

[ROCm] Eliminate redundant contiguous copy in MLA attention on ROCm MXFP4 (#25463)

2026-05-29 01:28:32 -07:00

afmoe.py

Fix fused_moe import for non-NPU devices (#25076)

2026-05-12 23:05:51 +03:00

apertus.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

arcee.py

Upgrade transformers==5.3.0 (#17784)

2026-03-18 13:50:43 -07:00

baichuan.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

bailing_moe_linear.py

[HotFix][Ling 2.6] Fix HybridLinearAttn dispatcher for Ling-2.6 (#26474)

2026-05-29 14:50:15 +08:00

bailing_moe_nextn.py

nextn subclass owns post_load_weights is_nextn (#24333)

2026-05-03 22:04:44 -07:00

bailing_moe.py

refactor(moe): centralize post-experts all-reduce skip predicate (#23748)

2026-04-26 20:29:59 -07:00

bert.py

[NPU] bugfix with Kimi-k2 and bge-reranker-v2 model (#17478)

2026-01-22 22:02:05 +08:00

chatglm.py

…

clip.py

VLM: add Conv2dLayer/Conv3dLayer to fix PyTorch 2.9.1 CuDNN Conv3d (#20282)

2026-03-15 19:17:44 +08:00

commandr.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

dbrx.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

deepseek_janus_pro.py

[AMD] Fix Janus-Pro crash and add Kimi-K2.5 nightly test (#18269)

2026-02-10 22:33:13 -08:00

deepseek_nextn.py

Support batch size > 1 when enable CP (#23269)

2026-05-27 14:11:17 -07:00

deepseek_ocr.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

deepseek_v2.py

[refactor] remove unused op_mlp (#26673)

2026-05-29 02:38:56 -07:00

deepseek_v4_nextn.py

refactor(dsv4): route MHC prenorm through DeepGEMM wrapper (#26238)

2026-05-27 17:45:45 -07:00

deepseek_v4.py

refactor(dsv4): route MHC prenorm through DeepGEMM wrapper (#26238)

2026-05-27 17:45:45 -07:00

deepseek_vl2.py

[VLM] Chunk-aware ViT encoding with per-image cache and lazy device transfer (#22038)

2026-04-04 16:55:17 +08:00

deepseek.py

[CPU] Add GPT-OSS model optimization for CPU (#16775)

2026-05-29 16:05:26 +08:00

dflash.py

[AMD] Enable DFLASH speculative decoding on ROCm (#22342)

2026-04-17 13:10:14 -07:00

dots_ocr.py

…

dots_vlm_vit.py

VLM: add Conv2dLayer/Conv3dLayer to fix PyTorch 2.9.1 CuDNN Conv3d (#20282)

2026-03-15 19:17:44 +08:00

dots_vlm.py

feat: support EPD disaggregation (#12263)

2025-12-14 22:30:08 +08:00

ernie4_eagle.py

update pre-commit config (#18860)

2026-02-16 00:18:31 +08:00

ernie4.py

[NPU] fix model ERNIE-4.5-21B-A3B-PT bias need 1D error (#26038)

2026-05-28 16:05:39 +08:00

ernie45_moe_vl.py

Upgrade transformers==5.3.0 (#17784)

2026-03-18 13:50:43 -07:00

ernie45_vl.py

[VLM] Support cos sin cache for Ernie4.5-VL (#19743)

2026-03-04 10:54:23 +08:00

exaone4.py

[Model] Add support for EXAONE-4.0 Model (#8205)

2026-01-27 14:08:24 +08:00

exaone_moe_mtp.py

[Model] Add K-EXAONE model support (#16294)

2026-01-30 20:01:14 +08:00

exaone_moe.py

refactor(moe): centralize post-experts all-reduce skip predicate (#23748)

2026-04-26 20:29:59 -07:00

exaone.py

fix: use get_rope_config() to support models without rope_parameters (#21135)

2026-03-26 11:22:12 -07:00

falcon_h1.py

[feat] Support extra_buffer in Mamba2-based models (#15829)

2026-05-26 16:03:29 +08:00

gemma2_reward.py

[Score API] Add return_pooled_hidden_states to Scoring API for SequenceClassification / RewardModel (#22427)

2026-04-15 14:58:56 -07:00

gemma2.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

gemma3_causal.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

gemma3_mm.py

feat(model_runner): remove pool/backend refs from ForwardBatch via ForwardContext (#25983)

2026-05-21 14:01:49 -07:00

gemma3n_audio.py

…

gemma3n_causal.py

Upgrade transformers==5.3.0 (#17784)

2026-03-18 13:50:43 -07:00

gemma3n_mm.py

Fix nightly VLM accuracy: gemma3n TP fixes + removal, latency thresholds (#19401)

2026-02-27 14:24:02 -08:00

gemma4_audio.py

[New Model] Gemma 4 (#21952)

2026-04-06 20:24:44 -07:00

gemma4_causal.py

[bug fix] Fix 3 issues when using Gemma4 MTP (#26026)

2026-05-23 03:16:47 -07:00

gemma4_mm.py

feat(model_runner): remove pool/backend refs from ForwardBatch via ForwardContext (#25983)

2026-05-21 14:01:49 -07:00

gemma4_mtp.py

[bug fix] Fix 3 issues when using Gemma4 MTP (#26026)

2026-05-23 03:16:47 -07:00

gemma4_vision.py

[New Model] Gemma 4 (#21952)

2026-04-06 20:24:44 -07:00

gemma.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

glm4_moe_lite_nextn.py

GLM-4.7-Flash: standalone MLA impl and MLA NextN/MTP (#26088)

2026-05-26 13:17:39 +08:00

glm4_moe_lite.py

[refactor] remove unused op_mlp (#26673)

2026-05-29 02:38:56 -07:00

glm4_moe_nextn.py

[Bug Fix] Align glm4_moe_nextn NPU MTP loading with qwen3 MTP (#25524)

2026-05-19 14:47:01 +01:00

glm4_moe.py

[refactor] remove unused op_mlp (#26673)

2026-05-29 02:38:56 -07:00

glm4.py

fix: use get_rope_config() to support models without rope_parameters (#21135)

2026-03-26 11:22:12 -07:00

glm4v_moe.py

[Bug Fix] GLM-V / GLM-OCR: field detection for transformers 5.x and MTP omission fix (#21134)

2026-03-23 13:19:48 -07:00

glm4v.py

[bugfix]GLM-4V model (#17122)

2026-04-01 10:37:40 +08:00

glm_ocr_nextn.py

[GLM-OCR] Support GLM-OCR Model (#17582)

2026-01-26 22:24:00 -08:00

glm_ocr.py

[Bug Fix] GLM-V / GLM-OCR: field detection for transformers 5.x and MTP omission fix (#21134)

2026-03-23 13:19:48 -07:00

glmasr.py

fix: change class name of GLM-ASR (#15772)

2025-12-26 00:15:16 +08:00

gpt2.py

update pre-commit config (#18860)

2026-02-16 00:18:31 +08:00

gpt_bigcode.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

gpt_j.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

gpt_oss.py

[CPU] Add GPT-OSS model optimization for CPU (#16775)

2026-05-29 16:05:26 +08:00

granite.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

granitemoe.py

Upgrade transformers==5.3.0 (#17784)

2026-03-18 13:50:43 -07:00

granitemoehybrid.py

[feat] Support extra_buffer in Mamba2-based models (#15829)

2026-05-26 16:03:29 +08:00

grok.py

fix(grok): adapt huihui-ai/grok-2 (#21522)

2026-04-07 10:04:41 +08:00

hunyuan_v3_nextn.py

support Hy3 preview (#23533)

2026-04-24 12:03:24 -07:00

hunyuan_v3.py

refactor(moe): centralize post-experts all-reduce skip predicate (#23748)

2026-04-26 20:29:59 -07:00

hunyuan.py

fix: use get_rope_config() to support models without rope_parameters (#21135)

2026-03-26 11:22:12 -07:00

idefics2.py

VLM: add Conv2dLayer/Conv3dLayer to fix PyTorch 2.9.1 CuDNN Conv3d (#20282)

2026-03-15 19:17:44 +08:00

internlm2_reward.py

[Score API] Add return_pooled_hidden_states to Scoring API for SequenceClassification / RewardModel (#22427)

2026-04-15 14:58:56 -07:00

internlm2.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

interns1.py

…

interns1pro.py

model: support interns1-pro (#18145)

2026-02-05 00:22:44 +08:00

interns2preview.py

Support Intern-S2-Preview (#24875)

2026-05-10 22:17:30 +08:00

internvl.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

iquest_loopcoder.py

fix: use get_rope_config() to support models without rope_parameters (#21135)

2026-03-26 11:22:12 -07:00

jet_nemotron.py

feat(model_runner): remove pool/backend refs from ForwardBatch via ForwardContext (#25983)

2026-05-21 14:01:49 -07:00

jet_vlm.py

…

kimi_k25_eagle3.py

[spec decoding] support kimi-k2.6-eagle3.1-mla draft (#26506)

2026-05-28 14:26:39 -07:00

kimi_k25.py

Enable Kimi-K2.5 piecewise CUDA graph (#26382)

2026-05-27 22:51:33 -07:00

kimi_linear.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

kimi_vl_moonvit.py

[XPU] Fix Device Assignment (#26257)

2026-05-29 09:38:11 +08:00

kimi_vl.py

[EPD][VLM] Support Kimi VL EPD (#22490)

2026-04-16 12:40:02 +08:00

laguna.py

[Model] Laguna-XS.2 Model Support (#24204)

2026-05-09 05:43:13 +08:00

lfm2_moe.py

Wire YARN rope_parameters through LFM2 and LFM2-MoE attention (#26187)

2026-05-26 23:00:22 +00:00

lfm2_vl.py

[VLM] fix LFM2-VL offline inference and GPU JPEG decode (#22448)

2026-04-15 09:13:25 +08:00

lfm2.py

Wire YARN rope_parameters through LFM2 and LFM2-MoE attention (#26187)

2026-05-26 23:00:22 +00:00

lightonocr.py

Support LightOnOCR-2-1B (#17806)

2026-01-29 23:03:41 +08:00

llada2.py

refactor(moe): centralize post-experts all-reduce skip predicate (#23748)

2026-04-26 20:29:59 -07:00

llama4.py

[Llama4] Use strided in-place fused QK RMSNorm to drop a redundant copy (#25089)

2026-05-15 10:33:08 -07:00

llama_classification.py

[Score API] Add SequenceClassification Model support (#22118)

2026-04-08 01:30:58 -07:00

llama_eagle3.py

[Refactor] Pass PP start_layer via model constructor instead of forward_batch.token_to_kv_pool (#25825)

2026-05-19 22:16:07 -07:00

llama_eagle.py

[Refactor] Pass PP start_layer via model constructor instead of forward_batch.token_to_kv_pool (#25825)

2026-05-19 22:16:07 -07:00

llama_embedding.py

…

llama_reward.py

[Score API] Add return_pooled_hidden_states to Scoring API for SequenceClassification / RewardModel (#22427)

2026-04-15 14:58:56 -07:00

llama.py

[NPU] fix model ERNIE-4.5-21B-A3B-PT bias need 1D error (#26038)

2026-05-28 16:05:39 +08:00

llava.py

perf: migrate Req token-id storage to array.array('q') in Scheduler (#25098)

2026-05-22 10:51:07 -07:00

llavavid.py

perf: migrate Req token-id storage to array.array('q') in Scheduler (#25098)

2026-05-22 10:51:07 -07:00

longcat_flash_nextn.py

[4/N] Quantization Refactor: AWQ schemes and Kernel call and weight init split (#21126)

2026-04-30 14:51:01 +03:00

longcat_flash.py

[4/N] Quantization Refactor: AWQ schemes and Kernel call and weight init split (#21126)

2026-04-30 14:51:01 +03:00

midashenglm.py

Upgrade transformers==5.3.0 (#17784)

2026-03-18 13:50:43 -07:00

mimo_audio.py

[Feature] Xiaomi MiMo-V2.5 day0 support (#23811)

2026-05-01 00:02:26 +08:00

mimo_mtp.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

mimo_v2_nextn.py

[Feature] Xiaomi MiMo-V2.5 day0 support (#23811)

2026-05-01 00:02:26 +08:00

mimo_v2.py

[refactor] remove unused op_mlp (#26673)

2026-05-29 02:38:56 -07:00

mimo_vl.py

[Feature] Xiaomi MiMo-V2.5 day0 support (#23811)

2026-05-01 00:02:26 +08:00

mimo.py

…

mindspore.py

feat(model_runner): remove pool/backend refs from ForwardBatch via ForwardContext (#25983)

2026-05-21 14:01:49 -07:00

minicpm3.py

fix: use get_rope_config() to support models without rope_parameters (#21135)

2026-03-26 11:22:12 -07:00

minicpm.py

fix: use get_rope_config() to support models without rope_parameters (#21135)

2026-03-26 11:22:12 -07:00

minicpmo.py

[XPU] Fix Device Assignment (#26257)

2026-05-29 09:38:11 +08:00

minicpmv_vit.py

[Model] Add MiniCPM-V 4.6 support (#24855)

2026-05-10 12:24:17 -04:00

minicpmv.py

[XPU] Fix Device Assignment (#26257)

2026-05-29 09:38:11 +08:00

minimax_m2.py

[refactor] remove unused op_mlp (#26673)

2026-05-29 02:38:56 -07:00

ministral3.py

Upgrade transformers==5.3.0 (#17784)

2026-03-18 13:50:43 -07:00

mistral_eagle.py

debug followup (#24058)

2026-04-29 23:03:27 +08:00

mistral_large_3_eagle.py

[CP] 1/N: Support MLA Prefill Context Parallel (#23292)

2026-05-23 03:07:09 -07:00

mistral_large_3.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

mistral.py

Bugfix (#24027)

2026-04-29 21:13:51 +08:00

mixtral_quant.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

mixtral.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

mllama4.py

Upgrade transformers==5.3.0 (#17784)

2026-03-18 13:50:43 -07:00

mllama.py

perf: migrate Req token-id storage to array.array('q') in Scheduler (#25098)

2026-05-22 10:51:07 -07:00

moss_vl.py

perf: migrate Req token-id storage to array.array('q') in Scheduler (#25098)

2026-05-22 10:51:07 -07:00

nano_nemotron_vl.py

[NemotronH] V3 Omni wrapper: WeightsMapper + config round-trip (#25023)

2026-05-26 20:34:05 +00:00

nemotron_h_mtp.py

Support NemotronHPuzzleForCausalLM (#24429)

2026-05-27 16:12:44 -07:00

nemotron_h.py

Support NemotronHPuzzleForCausalLM (#24429)

2026-05-27 16:12:44 -07:00

nemotron_nas.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

nvila_lite.py

…

nvila.py

…

olmo2.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

olmo.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

olmoe.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

opt.py

update pre-commit config (#18860)

2026-02-16 00:18:31 +08:00

orion.py

fix: use get_rope_config() to support models without rope_parameters (#21135)

2026-03-26 11:22:12 -07:00

paddleocr_vl.py

VLM: add Conv2dLayer/Conv3dLayer to fix PyTorch 2.9.1 CuDNN Conv3d (#20282)

2026-03-15 19:17:44 +08:00

parakeet.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

persimmon.py

Upgrade transformers==5.3.0 (#17784)

2026-03-18 13:50:43 -07:00

phi3_small.py

Upgrade transformers==5.3.0 (#17784)

2026-03-18 13:50:43 -07:00

phi4mm_audio.py

…

phi4mm_utils.py

…

phi4mm.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

phi.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

phimoe.py

Upgrade transformers==5.3.0 (#17784)

2026-03-18 13:50:43 -07:00

pixtral.py

Bugfix (#24027)

2026-04-29 21:13:51 +08:00

points_v15_chat.py

…

qwen2_5_vl.py

[NPU] Convert cu_window_seqlens to CPU for npu_flush_attention_unpad operator (#20328)

2026-03-12 09:08:43 +08:00

qwen2_audio.py

update pre-commit config (#18860)

2026-02-16 00:18:31 +08:00

qwen2_classification.py

[Score API] Add SequenceClassification Model support (#22118)

2026-04-08 01:30:58 -07:00

qwen2_eagle.py

[Refactor] Pass PP start_layer via model constructor instead of forward_batch.token_to_kv_pool (#25825)

2026-05-19 22:16:07 -07:00

qwen2_moe.py

feat: Support flashinfer_cutedsl MoE runner with flashinfer alltoall backend (#22669)

2026-05-20 00:36:26 -07:00

qwen2_rm.py

[Score API] Add return_pooled_hidden_states to Scoring API for SequenceClassification / RewardModel (#22427)

2026-04-15 14:58:56 -07:00

qwen2_vl.py

VLM: add Conv2dLayer/Conv3dLayer to fix PyTorch 2.9.1 CuDNN Conv3d (#20282)

2026-03-15 19:17:44 +08:00

qwen2.py

[Refactor] Pass PP start_layer via model constructor instead of forward_batch.token_to_kv_pool (#25825)

2026-05-19 22:16:07 -07:00

qwen3_5_mtp.py

[BugFix] Resolve adaptive speculative decoding conflicts for Qwen3.5 (hybrid GDN) (#23331)

2026-05-19 15:09:49 -07:00

qwen3_5.py

[NPU]use triton split_qkvgate_gemma_rmsnorm_rope for Qwen3.5 and Qwen3_next (#23925)

2026-05-20 20:22:10 +08:00

qwen3_asr.py

model: support qwen3-asr (#22073)

2026-04-07 13:27:05 +08:00

qwen3_classification.py

[Model] Qwen3ForPooledOutput: forward get_input_embeddings to inner model (#23434)

2026-04-29 12:25:06 -07:00

qwen3_moe.py

[refactor] remove unused op_mlp (#26673)

2026-05-29 02:38:56 -07:00

qwen3_next_mtp.py

[BugFix] Resolve adaptive speculative decoding conflicts for Qwen3.5 (hybrid GDN) (#23331)

2026-05-19 15:09:49 -07:00

qwen3_next.py

[NPU]use triton split_qkvgate_gemma_rmsnorm_rope for Qwen3.5 and Qwen3_next (#23925)

2026-05-20 20:22:10 +08:00

qwen3_omni_moe.py

[CPU] Add support for Qwen3-vl and Qwen3-omni (#12662)

2026-05-27 08:56:09 +08:00

qwen3_rm.py

[Model] Add Qwen3ForRewardModel and fix Qwen3ForSequenceClassification (#17992)

2026-02-16 19:44:41 +08:00

qwen3_vl_moe.py

[BugFix][EPD]Fix Qwen3VLMoe encoder-only AttributeError (#25759)

2026-05-21 11:07:21 +08:00

qwen3_vl.py

[CPU] Add support for Qwen3-vl and Qwen3-omni (#12662)

2026-05-27 08:56:09 +08:00

qwen3.py

feat(model_runner): remove pool/backend refs from ForwardBatch via ForwardContext (#25983)

2026-05-21 14:01:49 -07:00

qwen.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

radio.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

registry.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

roberta.py

…

sarashina2_vision.py

…

sarvam_moe.py

Route concat MLA to JIT and remove unused downcast (#25843)

2026-05-23 14:30:43 +08:00

sdar_moe.py

refactor(moe): centralize post-experts all-reduce skip predicate (#23748)

2026-04-26 20:29:59 -07:00

sdar.py

Feature/sdar support (#19044)

2026-02-19 21:58:15 -08:00

siglip2.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

siglip.py

[CPU] Add frontend support for Gemma (#12590)

2026-03-18 23:02:26 -07:00

solar.py

fix: use get_rope_config() to support models without rope_parameters (#21135)

2026-03-26 11:22:12 -07:00

stablelm.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

starcoder2.py

Upgrade transformers==5.3.0 (#17784)

2026-03-18 13:50:43 -07:00

step3_vl_10b.py

[VLM] Chunk-aware ViT encoding with per-image cache and lazy device transfer (#22038)

2026-04-04 16:55:17 +08:00

step3_vl.py

fix: use get_rope_config() to support models without rope_parameters (#21135)

2026-03-26 11:22:12 -07:00

step3p5_mtp.py

[misc] refine outdated comments for chain-style multi-layer MTP (#22996)

2026-04-16 14:49:43 -07:00

step3p5.py

model: support Step-3.7-Flash (#26565)

2026-05-29 08:00:54 +08:00

step3p7.py

model: support Step-3.7-Flash (#26565)

2026-05-29 08:00:54 +08:00

teleflm.py

…

torch_native_llama.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

transformers.py

[XPU] Fix Device Assignment (#26257)

2026-05-29 09:38:11 +08:00

utils.py

feat(model_runner): remove pool/backend refs from ForwardBatch via ForwardContext (#25983)

2026-05-21 14:01:49 -07:00

voxtral.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

whisper.py

perf: migrate Req token-id storage to array.array('q') in Scheduler (#25098)

2026-05-22 10:51:07 -07:00

xverse_moe.py

refactor(moe): de-duplicate triton MoE runner path into shared helpers (#23019)

2026-04-17 17:05:13 -07:00

xverse.py

chore: add vLLM SPDX copyright headers to ported files (#25182)

2026-05-13 15:17:30 -07:00

yivl.py

Upgrade transformers from 5.5.4 to 5.6.0 (#23525)

2026-04-26 22:33:54 -07:00