Commits · 1838cd4860cf96972812c80a1cfc788f9ba3584b · OpenDAS / vllm_cscc

"tests/benchmarks/sweep/__init__.py" did not exist on "e575df33b1963212f79b2efddd45d53225fa5c45"

04 Oct, 2025 9 commits

Revert "Add batch invariant kernel override for FlashInfer backend [2/n]" (#26220) · 1838cd48
Cyrus Leung authored Oct 04, 2025

1838cd48
[Model] Support nested structures for TensorSchema (#26212) · 44ea8513
Cyrus Leung authored Oct 04, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
44ea8513

Support expert parallel in Transformers backend (#26162) · d3d649ef

Harry Mellor authored Oct 04, 2025


Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>

d3d649ef

[V1] [Hybrid] Mamba2 Automatic Prefix Caching (#25752) · ea507c3a

Stan Wozniak authored Oct 04, 2025


Signed-off-by: Stanislaw Wozniak <stw@zurich.ibm.com>
Signed-off-by: Thomas Ortner <boh@zurich.ibm.com>
Signed-off-by: Thomas Parnell <tpa@zurich.ibm.com>
Co-authored-by: Thomas Ortner <boh@zurich.ibm.com>
Co-authored-by: Thomas Parnell <tpa@zurich.ibm.com>

ea507c3a

[cpu][perf] Accelerate unquantized-linear for AArch64 through oneDNN/ACL and... · 9705fba7

Fadi Arafeh authored Oct 04, 2025


[cpu][perf] Accelerate unquantized-linear for AArch64 through oneDNN/ACL and weight prepack (#25948)
Signed-off-by: Fadi Arafeh <fadi.arafeh@arm.com>
Co-authored-by: Li, Jiang <jiang1.li@intel.com>

9705fba7

Add batch invariant kernel override for FlashInfer backend [2/n] (#25769) · 2f7dbc9b

Bram Wasti authored Oct 03, 2025


Signed-off-by: Bram Wasti <bwasti@meta.com>
Signed-off-by: Bram Wasti <bwasti@fb.com>
Co-authored-by: Wentao Ye <44945378+yewentao256@users.noreply.github.com>

2f7dbc9b

[Bugfix] Fix qwen3 vl dummy data generation with overrides (#26193) · 67bc0c00
Roger Wang authored Oct 03, 2025
```
Signed-off-by: Roger Wang <hey@rogerw.io>
```
67bc0c00
Fix issue of using only the part of video frame [Nemotron Nano] (#26186) · 5a05f266
Eugene Khvedchenya authored Oct 04, 2025
```
Signed-off-by: Eugene Khvedchenia <ekhvedchenia@nvidia.com>
```
5a05f266

[GPTOSS][DP/EP][Marlin] Enable GPTOSS DP/EP using Marlin kernels (#25488) · 7ef40bb9

Varun Sundar Rabindranath authored Oct 03, 2025


Signed-off-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Co-authored-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Co-authored-by: mgoin <mgoin64@gmail.com>

7ef40bb9

03 Oct, 2025 15 commits
- [Refactor] Optimize FP8 MOE Backend Choice and Log (#26044) · c1ffcb55
  Wentao Ye authored Oct 03, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  c1ffcb55
- [Quantization/NVFP4] Speed up TRTLLM NVFP4 MOE weight loading and fix K/V... · a2691733
  Pavani Majety authored Oct 03, 2025
```
[Quantization/NVFP4] Speed up TRTLLM NVFP4 MOE weight loading and fix K/V scale loading for MLA Attn (#25968)
Signed-off-by: Pavani Majety <pmajety@nvidia.com>
```
  a2691733
- [BugFix][QWEN-VL]fix wrong apply_rotary_emb_torch selection introduced by #24642 (#26123) · dd96465f
  Chendi.Xue authored Oct 03, 2025
```
Signed-off-by: Chendi Xue <Chendi.Xue@intel.com>
Signed-off-by: Chendi.Xue <chendi.xue@intel.com>
Co-authored-by: Roger Wang <hey@rogerw.io>
```
  dd96465f
- [Model] Supplement to PR 24862: Pass param prefix to LLMHead (#25805) · cbf92219
  whx authored Oct 03, 2025
```
Signed-off-by: whx-sjtu <2952154980@qq.com>
```
  cbf92219
- [backends][short_conv] CUDA graph piecewise edits (#24215) · 5f42fc53
  Paul Pak authored Oct 03, 2025
```
Signed-off-by: Paul Pak <paulpak58@gmail.com>
```
  5f42fc53
- [Doc] Fixed shape description for fused_batched_moe.py (#25668) · aa5053e3
  Egor authored Oct 03, 2025
```
Signed-off-by: Egor <e.a.krivov@gmail.com>
```
  aa5053e3
- [Multi Modal] Configurable MM Profiling (#25631) · 79aa2446
  Wenlong Wang authored Oct 03, 2025
```
Signed-off-by: wwl2755 <wangwenlong2755@gmail.com>
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  79aa2446
- [Model] Use `merge_by_field_config` for MM models (InternVL family) (#26153) · f9a8084e
  Cyrus Leung authored Oct 03, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  f9a8084e
- [Input] Remove unused `prompt` field (#26097) · 0ad9951c
  Cyrus Leung authored Oct 03, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  0ad9951c
- [Misc] Remove typing.List (#26150) · 8c911718
  Varun Sundar Rabindranath authored Oct 03, 2025
```
Signed-off-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
Co-authored-by: Varun Sundar Rabindranath <vsundarr@redhat.com>
```
  8c911718
- `FusedMoE` support for the Transformers backend (#22650) · 10d76548
  Harry Mellor authored Oct 03, 2025
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
```
  10d76548
- [Model] Use `merge_by_field_config` for MM models (G) (#26117) · 39b643dc
  Cyrus Leung authored Oct 03, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  39b643dc
- [Bugfix] Fix import `gemm_afp4wfp4` failure on AMD (#26068) · 711f4856
  Zhewen Li authored Oct 02, 2025
```
Signed-off-by: zhewenli <zhewenli@meta.com>
```
  711f4856
- [ROCm] [VL] [Bugfix] Fix vit flash attn dispatcher logic for ROCm (#26104) · 9c5ee91b
  TJian authored Oct 02, 2025
```
Signed-off-by: tjtanaa <tunjian.tan@embeddedllm.com>
```
  9c5ee91b
- [Attention] Move Backend enum into registry (#25893) · 2aaa4238
  Matthew Bonanni authored Oct 02, 2025
```
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
```
  2aaa4238
02 Oct, 2025 6 commits

Fix MTP with deepep_low_latency (#25904) · 13cdc021
Matthew Bonanni authored Oct 02, 2025
```
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
```
13cdc021

[Perf] Fix and reapply move apply w8a8 block fp8 linear to class (#25696) · 502640c3

ElizaWszola authored Oct 02, 2025


Signed-off-by: ElizaWszola <ewszola@redhat.com>
Signed-off-by: ElizaWszola <elizaw.9289@gmail.com>
Signed-off-by: Luka Govedič <lgovedic@redhat.com>
Signed-off-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>
Co-authored-by: Luka Govedič <lgovedic@redhat.com>

502640c3

[Deepseek v3.2] Support indexer prefill chunking (#25999) · 1e50f1be
Chen Zhang authored Oct 02, 2025
```
Signed-off-by: Chen Zhang <zhangch99@outlook.com>
```
1e50f1be
[Qwen][ROCm] Flash Attention Rotary Embeddings (#24642) · 5e4a8223
vllmellm authored Oct 02, 2025
```
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
```
5e4a8223
[Model] Use `merge_by_field_config` for MM models (D-F) (#26076) · cc253b73
Cyrus Leung authored Oct 02, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
cc253b73
[Model] Use `merge_by_field_config` for MM models (A-C) (#26073) · 7d6fb905
Cyrus Leung authored Oct 02, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
7d6fb905

01 Oct, 2025 7 commits
- Support RL online quantization with torchao (#23014) · c3124680
  Jerry Zhang authored Oct 01, 2025
```
Signed-off-by: Jerry Zhang <jerryzh168@gmail.com>
```
  c3124680
- [Misc] Factor out common `_apply_feature_select_strategy` (#26003) · 1405f0c7
  Cyrus Leung authored Oct 01, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  1405f0c7
- [BugFix][MM] Fix Nonetype error when video is cache in qwen2.5-omni-thinker (#26004) · 84d57342
  Wenlong Wang authored Oct 01, 2025
```
Signed-off-by: wwl2755 <wangwenlong2755@gmail.com>
```
  84d57342
- Update to Transformers `v4.56.2` (#24638) · 2a69ab48
  Harry Mellor authored Oct 01, 2025
```
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>
```
  2a69ab48
- [BugFix] Fix default kv-cache-dtype default for DeepseekV3.2 (#25988) · 8d7da92f
  Lucas Wilkinson authored Oct 01, 2025
```
Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
```
  8d7da92f
- [MM] Add text-only mode for Qwen3-VL (#26000) · 66bca9b8
  Roger Wang authored Sep 30, 2025
  
  66bca9b8
- [Log] Optimize Log for FP8MOE (#25709) · 12449488
  Wentao Ye authored Sep 30, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
  12449488
30 Sep, 2025 3 commits

Add explicit pooling classes for the Transformers backend (#25322) · a388252a

Harry Mellor authored Sep 30, 2025


Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Co-authored-by: Isotr0py <mozf@mail2.sysu.edu.cn>

a388252a

[Llama4] [multimodal] Fix misplaced dtype cast of `cos_sin_cache` in... · 43b752c3

cjackal authored Oct 01, 2025


[Llama4] [multimodal] Fix misplaced dtype cast of `cos_sin_cache` in `Llama4VisionRotaryEmbedding` (#25889)
Signed-off-by: cjackal <44624812+cjackal@users.noreply.github.com>

43b752c3

[Bug] Fix AttributeError: 'QKVParallelLinear' object has no attribute 'orig_dtype' (#25958) · e6a226ef
Wentao Ye authored Sep 30, 2025
```
Signed-off-by: yewentao256 <zhyanwentao@126.com>
```
e6a226ef