Commits · 5e4e0e51f4fbb6d6fbf7a94feef07bd3617bfa5f · OpenDAS / vllm_cscc

23 Jan, 2026 1 commit

[torch.compile] Compile `CustomOp.forward_native` for `SiluAndMul` and... · 5e4e0e51

Luka Govedič authored Jan 22, 2026


[torch.compile] Compile `CustomOp.forward_native` for `SiluAndMul` and `QuantFP8` to avoid raw torch ops inside opaque custom ops (#32806)
Signed-off-by: Luka Govedič <lgovedic@redhat.com>
Signed-off-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
Co-authored-by: Michael Goin <mgoin64@gmail.com>

5e4e0e51

22 Jan, 2026 11 commits
- [MoE Refactor] Move `select_experts` from `FusedMoEQuantMethod` -> `FusedMoE` (#31996) · dc917cce
  bnellnm authored Jan 22, 2026
```
Signed-off-by: Bill Nell <bnell@redhat.com>
```
  dc917cce
- [BugFix] Fix invalid flashinfer_fused_moe_blockscale_fp8 op registration (#32855) · fc56f4a0
  Fadi Arafeh authored Jan 22, 2026
```
Signed-off-by: Fadi Arafeh <fadi.arafeh@arm.com>
```
  fc56f4a0
- [Perf] Create TMA-aligned input scale tensor for DeepGemm on Hopper (#32619) · d08b356e
  Xin Yang authored Jan 22, 2026
```
Signed-off-by: Xin Yang <xyangx@amazon.com>
```
  d08b356e
- Add llmcompressor fp8 kv-cache quant (per-tensor and per-attn_head) (#30141) · 44f08af3
  Eldar Kurtić authored Jan 22, 2026
```
Signed-off-by: Eldar Kurtic <8884008+eldarkurtic@users.noreply.github.com>
Signed-off-by: eldarkurtic <8884008+eldarkurtic@users.noreply.github.com>
```
  44f08af3
- [Bugfix][Attention] Explicitly report support for kv_cache_dtype bfloat16 (#32795) · 955b43a5
  Matthew Bonanni authored Jan 22, 2026
```
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
```
  955b43a5
- Support bge-m3 sparse embeddings and colbert embeddings (#14526) · ff365eea
  Maximilien de Bayser authored Jan 22, 2026
```
Signed-off-by: Max de Bayser <mbayser@br.ibm.com>
Signed-off-by: Max de Bayser <maxdebayser@gmail.com>
```
  ff365eea
- [torch.compile] Improve Cold Start for MoEs (#32805) · 654a71fc
  Richard Zou authored Jan 22, 2026
```
Signed-off-by: Richard Zou <zou3519@gmail.com>
```
  654a71fc
- [AMD][ROCm] MoRI EP: a high-performance all2all backend (#28664) · 49a12622
  Alex Sun authored Jan 22, 2026
```
Signed-off-by: Alex Sun <alex.s@amd.com>
```
  49a12622
- [Model] Extend `collect_children` and `no_init_weights` contexts (#32757) · 2b8a38b6
  Cyrus Leung authored Jan 22, 2026
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  2b8a38b6
- [ROCm][CI] Fix AITER test flakiness by using explicit attention backend (#32346) · eb1629da
  Andreas Karatzas authored Jan 21, 2026
```
Signed-off-by: Andreas Karatzas <akaratza@amd.com>
Signed-off-by: Matthew Wong <Matthew.Wong2@amd.com>
Co-authored-by: Matthew Wong <Matthew.Wong2@amd.com>
```
  eb1629da
- [Llama.py -> mistral.py] Extract mistral-only relevant code into separate file (#32780) · 1579c9b5
  Patrick von Platen authored Jan 22, 2026
```
Signed-off-by: Patrick von Platen <patrick.v.platen@gmail.com>
```
  1579c9b5
21 Jan, 2026 17 commits
- [Kernel] Add topk_sigmoid kernel (#31246) · 63227acc
  Xin Yang authored Jan 21, 2026
```
Signed-off-by: Xin Yang <xyangx@amazon.com>
```
  63227acc
- [ROCm] fix import for on_gfx9 (#32783) · 180fba65
  Divakar Verma authored Jan 21, 2026
```
Signed-off-by: Divakar Verma <divakar.verma@amd.com>
```
  180fba65
- Bump Flashinfer to v0.6.1 (#30993) · 808d6fd7
  elvischenv authored Jan 22, 2026
```
Signed-off-by: elvischenv <219235043+elvischenv@users.noreply.github.com>
```
  808d6fd7
- [PluggableLayer][1/N] Define PluggableLayer (Fix ci) (#32744) · 1861ae8a
  whx authored Jan 22, 2026
```
Signed-off-by: whx-sjtu <2952154980@qq.com>
```
  1861ae8a
- [Quantization][Deprecation] Remove RTN (#32697) · 4e31b7f2
  Robert Shaw authored Jan 21, 2026
```
Signed-off-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Robert Shaw <robshaw@redhat.com>
```
  4e31b7f2
- [ROCm][Deepseekv3.2] Refactor Sparse Indexer as CustomOp (#29287) · 6c20e89c
  Pleaplusone authored Jan 21, 2026
```
Signed-off-by: ganyi <ygan@amd.com>
```
  6c20e89c
- [Quantization][Deprecation] Deprecate HQQ (#32681) · 85f55c94
  Robert Shaw authored Jan 21, 2026
```
Signed-off-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Robert Shaw <robshaw@redhat.com>
```
  85f55c94
- [Quantization][Deprecation] Remove `DeepSpeedFp8` (#32679) · cea3c754
  Robert Shaw authored Jan 21, 2026
```
Signed-off-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Robert Shaw <robshaw@redhat.com>
```
  cea3c754
- [MoE Refactor] Oracle Select FP8+NVFP4 Kernels In Priority (#32414) · 42135d68
  Robert Shaw authored Jan 21, 2026
  
  42135d68
- [bugfix] Aria model (#32727) · e14467be
  Divakar Verma authored Jan 21, 2026
```
Signed-off-by: Divakar Verma <divakar.verma@amd.com>
```
  e14467be
- [Model] Add Eagle2.5-8B Vision-Language Model support (#32456) · 7727ce35
  Kim Hee Su authored Jan 21, 2026
```
Signed-off-by: kimheesu <wlskaka4@gmail.com>
```
  7727ce35
- [Bugfix] Support HF sharded weights for Mistral3/Pixtral models (#32673) · f23fb5a7
  RickyChen / 陳昭儒 authored Jan 21, 2026
```
Signed-off-by: ricky-chaoju <ricky.chen@infinirc.com>
Signed-off-by: vllm-dev <ricky.chen@infinirc.com>
```
  f23fb5a7
- [Bugfix] Fix Nemotron-Nano-v2-vlm static resolution (#32682) · 27ca95b3
  Netanel Haber authored Jan 21, 2026
```
Signed-off-by: Netanel Haber <58652339+netanel-haber@users.noreply.github.com>
```
  27ca95b3
- Added qwen3 vision language moe support for speculative decoding (#32048) · 7ab80a8e
  shanjiaz authored Jan 20, 2026
```
Signed-off-by: shanjiaz <zsjwpianpian@gmail.com>
Signed-off-by: shanjiaz <43143795+shanjiaz@users.noreply.github.com>
```
  7ab80a8e
- Enable Eagle3 speculative decoding for Pixtral (LlavaForConditionalGeneration) (#32542) · 0900cedb
  gopalsarda authored Jan 20, 2026
```
Signed-off-by: gopalsarda <gopal.sarda@servicenow.com>
```
  0900cedb
- [Bugfix] Fix Granite Vision / Don't use Siglip Pooling Head Nested Models by Default (#32299) · 27b81e01
  Alex Brooks authored Jan 20, 2026
```
Signed-off-by: Alex-Brooks <Alex.Brooks@ibm.com>
```
  27b81e01
- Revert "[PluggableLayer][1/N] Define PluggableLayer" (#32725) · c78ee240
  Robert Shaw authored Jan 20, 2026
  
  c78ee240
20 Jan, 2026 11 commits
- fp8 online quant: split out Fp8OnlineLinearMethod (#32189) · d2389c12
  Vasiliy Kuznetsov authored Jan 20, 2026
  
  d2389c12
- [Bugfix] Suppress log on non-ROCm platform (#32703) · 9b67338b
  TJian authored Jan 21, 2026
```
Signed-off-by: tjtanaa <tunjian.tan@embeddedllm.com>
```
  9b67338b
- [5/N] Initialize MM components in context managers (Q-Z) (#32695) · 193069d1
  Cyrus Leung authored Jan 21, 2026
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  193069d1
- [Bugfix] fix the ima issue of qwen-vit (#32687) · 04a9e064
  JJJYmmm authored Jan 21, 2026
```
Signed-off-by: JJJYmmm <92386084+JJJYmmm@users.noreply.github.com>
```
  04a9e064
- [PluggableLayer][1/N] Define PluggableLayer (#32331) · 4ca62a0d
  whx authored Jan 21, 2026
```
Signed-off-by: whx-sjtu <2952154980@qq.com>
```
  4ca62a0d
- [4/N] Initialize MM components in context managers (M-P) (#32663) · fda3f03e
  Cyrus Leung authored Jan 20, 2026
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  fda3f03e
- [Bugfix] Fix the fp8_mqa_logits dim mismatch (#32652) · c4e5bdf6
  Chauncey authored Jan 20, 2026
```
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
```
  c4e5bdf6
- [3/N] Initialize MM components in context managers (I-L) (#32650) · 7f1bcd18
  Cyrus Leung authored Jan 20, 2026
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  7f1bcd18
- [2/N] Initialize MM components in context managers (E-H) (#32641) · e1a34c3a
  Cyrus Leung authored Jan 20, 2026
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  e1a34c3a
- [Refactor] Make FP8 Linear Ops use kernel abstraction (#27814) · 148117ea
  vllmellm authored Jan 20, 2026
```
Signed-off-by: vllmellm <vllm.ellm@embeddedllm.com>
```
  148117ea
- [1/N] Initialize MM components in context managers (A-D) (#32632) · b75e85de
  Cyrus Leung authored Jan 20, 2026
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  b75e85de