Commits · b0ccfc565ac263b4097175b3ba33ad1d5dc5df8a · OpenDAS / vllm_cscc · GitLab

14 Feb, 2025 8 commits
- [Bugfix][V1] GPUModelRunner._update_states should return True when there is a... · b0ccfc56
  Kero Liang authored Feb 14, 2025
```
[Bugfix][V1] GPUModelRunner._update_states should return True when there is a finished request in batch (#13126)
```
  b0ccfc56
- [ROCm][V1] Add intial ROCm support to V1 (#12790) · ba59b78a
  Sage Moore authored Feb 13, 2025
  
  ba59b78a
- [V1] LoRA - Enable Serving Usecase (#12883) · cbc40128
  Varun Sundar Rabindranath authored Feb 14, 2025
```
Signed-off-by: Varun Sundar Rabindranath <varun@neuralmagic.com>
Co-authored-by: Varun Sundar Rabindranath <varun@neuralmagic.com>
```
  cbc40128
- Expand MLA to support most types of quantization (#13181) · f0b2da72
  Michael Goin authored Feb 14, 2025
  
  f0b2da72
- [Misc] Remove redundant statements in scheduler.py (#13229) · 40932d7a
  Wang Ran (汪然) authored Feb 14, 2025
  
  40932d7a
- [V1] Consolidate MM cache size to vllm.envs (#13239) · dd5ede44
  Roger Wang authored Feb 13, 2025
  
  dd5ede44
- [ROCm] Avoid using the default stream on ROCm (#13238) · 41088695
  Gregory Shtrasberg authored Feb 13, 2025
```
Signed-off-by: Gregory Shtrasberg <Gregory.Shtrasberg@amd.com>
```
  41088695
- [Kernel][Bugfix] Refactor and Fix CUTLASS 2:4 Sparse Kernels (#13198) · c1e37bf7
  Tyler Michael Smith authored Feb 13, 2025
```
Signed-off-by: Tyler Michael Smith <tyler@neuralmagic.com>
```
  c1e37bf7
13 Feb, 2025 17 commits
- Optimize moe_align_block_size for deepseek_v3 (#12850) · 2344192a
  Michael Goin authored Feb 13, 2025
```
Signed-off-by: mgoin <mgoin64@gmail.com>
```
  2344192a
- [Frontend] Add `/v1/audio/transcriptions` OpenAI API endpoint (#12909) · d84cef76
  Nicolò Lucchesi authored Feb 13, 2025
  
  d84cef76
- [Bugfix] Missing Content Type returns 500 Internal Server Error (#13193) · 37dfa600
  Vaibhav Jain authored Feb 13, 2025
  
  37dfa600
- [VLM] Separate text-only and vision variants of the same model architecture (#13157) · 1bc3b5e7
  Cyrus Leung authored Feb 13, 2025
  
  1bc3b5e7
- [Misc] Qwen2.5-VL Optimization (#13155) · 02ed8a1f
  燃 authored Feb 13, 2025
  
  02ed8a1f
- [V1][Core] Add worker_base for v1 worker (#12816) · 2092a6fa
  Aoyu authored Feb 13, 2025
```
Signed-off-by: Aoyu <aoyuzhan@amazon.com>
Signed-off-by: youkaichao <youkaichao@gmail.com>
Co-authored-by: Aoyu <aoyuzhan@amazon.com>
Co-authored-by: youkaichao <youkaichao@gmail.com>
```
  2092a6fa
- [VLM] Merged multi-modal processor for Molmo (#12966) · c9d3ecf0
  Cyrus Leung authored Feb 13, 2025
  
  c9d3ecf0
- [V1] Clarify input processing and multimodal feature caching logic (#13211) · fdcf64d3
  Roger Wang authored Feb 13, 2025
  
  fdcf64d3
- [Frontend] Pass pre-created socket to uvicorn (#13113) · 578087e5
  Russell Bryant authored Feb 13, 2025
  
  578087e5
- [VLM] Remove input processor from clip and siglip (#13165) · fa253f1a
  Isotr0py authored Feb 13, 2025
  
  fa253f1a
- [V1][core] Implement pipeline parallel on Ray (#12996) · 9605c125
  Rui Qiao authored Feb 13, 2025
  
  9605c125
- Allow Unsloth Dynamic 4bit BnB quants to work (#12974) · cb944d58
  Daniel Han authored Feb 12, 2025
  
  cb944d58
- [Frontend] Move CLI code into vllm.cmd package (#12971) · d46d490c
  Russell Bryant authored Feb 13, 2025
  
  d46d490c
- [Bugfix] deepseek_r1_reasoning_parser put reason content in wrong field in... · 04f50ad9
  LikeSundayLikeRain authored Feb 13, 2025
```
[Bugfix] deepseek_r1_reasoning_parser put reason content in wrong field in certain edge case (#13097)
```
  04f50ad9
- Simplify logic of locating CUDART so file path (#13203) · 009439ca
  Lu Fang authored Feb 12, 2025
```
Signed-off-by: Lu Fang <lufang@fb.com>
```
  009439ca
- [VLM] Implement merged multimodal processor for Mllama (#11427) · bc55d130
  Isotr0py authored Feb 13, 2025
  
  bc55d130
- [NVIDIA] Support nvfp4 quantization (#12784) · 4fc5c23b
  Kaixi Hou authored Feb 12, 2025
  
  4fc5c23b
12 Feb, 2025 13 commits
- [V1][Bugfix] Copy encoder input ids to fix set iteration issue during VLM abort (#13173) · 4c0d93f4
  Murali Andoorveedu authored Feb 12, 2025
```
Signed-off-by: andoorve <37849411+andoorve@users.noreply.github.com>
```
  4c0d93f4
- [Bugfix] Allow fallback to AWQ from AWQMarlin at per-layer granularity (#13119) · 09972e71
  Michael Goin authored Feb 12, 2025
  
  09972e71
- [CORE] [QUANT] Support for GPTQModel's `dynamic` quantization per module override/control (#7086) · 36a08630
  Qubitium-ModelCloud authored Feb 13, 2025
  
  36a08630
- Introduce VLLM_CUDART_SO_PATH to allow users specify the .so path (#12998) · 042c3419
  Lu Fang authored Feb 12, 2025
```
Signed-off-by: Lu Fang <lufang@fb.com>
```
  042c3419
- [Misc] Delete unused LoRA modules (#13151) · 82cabf53
  Jee Jee Li authored Feb 13, 2025
  
  82cabf53
- [Frontend] Generate valid tool call IDs when using `tokenizer-mode=mistral` (#12332) · 314cfade
  Rafael Vasquez authored Feb 12, 2025
  
  314cfade
- [Bugfix] Fix num video tokens calculation for Qwen2-VL (#13148) · 985b4a2b
  Cyrus Leung authored Feb 12, 2025
```
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
```
  985b4a2b
- [Bug] [V1] Try fetching stop_reason from EngineOutput before checking the request (#13108) · f4d97e4f
  bnellnm authored Feb 12, 2025
  
  f4d97e4f
- [Misc] AMD Build Improvements (#12923) · f1042e86
  Shiyan Deng authored Feb 12, 2025
  
  f1042e86
- Further reduce the HTTP calls to huggingface.co (#13107) · 7c4033ac
  Maximilien de Bayser authored Feb 12, 2025
  
  7c4033ac
- [Neuron][Kernel] Support Longer Sequences in NKI-based Flash PagedAttention... · e92694b6
  Lingfan Yu authored Feb 11, 2025
```
[Neuron][Kernel] Support Longer Sequences in NKI-based Flash PagedAttention and Improve Efficiency (#12921)
Signed-off-by: Lingfan Yu <lingfany@amazon.com>
```
  e92694b6
- [Model] IBM/NASA Prithvi Geospatial model (#12830) · 974dfd49
  Christian Pinto authored Feb 12, 2025
  
  974dfd49
- [RFC][vllm-API] Support tokenizer registry for customized tokenizer in vLLM (#12518) · 3ee696a6
  Keyun Tong authored Feb 11, 2025
```
Signed-off-by: Keyun Tong <tongkeyun@gmail.com>
```
  3ee696a6
11 Feb, 2025 2 commits
- [Bugfix] Guided decoding falls back to outlines when fails to import xgrammar (#12976) · 14ecab5b
  Yuan Tang authored Feb 11, 2025
```
Signed-off-by: Yuan Tang <terrytangyuan@gmail.com>
```
  14ecab5b
- [CI/Build][Bugfix] Fix CPU backend default threads num (#13077) · 565c1efa
  Li, Jiang authored Feb 12, 2025
  
  565c1efa