Skip to content

Pull requests: vllm-project/vllm

New pull request New
472 Open 6,323 Closed
472 Open 6,323 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

[Bugfix] Fix DeepSeek MTP crash when using TP1ModelRunner with CUDA graph due to shape mismatch speculative-decoding
#14237 opened Mar 4, 2025 by pyc96 Loading…
1
[V1][Bugfix] Do not reset prefix caching metrics ready ONLY add when PR is ready to merge/full CI is needed v1
#14235 opened Mar 4, 2025 by comaniac Loading…
2
Adjusting invalid k values in top-k selection beyond vocabulary limits v1
#14234 opened Mar 4, 2025 by dashanji Loading…
1
2
Torchao documentation Improvements or additions to documentation
#14231 opened Mar 4, 2025 by drisspg • Draft
1
Use getattr for hidden_act and hidden_activation in Gemma models
#14230 opened Mar 4, 2025 by richardsliu Loading…
1
[CI] Make UT cases in test_comm_ops.py compatible with more devices
#14229 opened Mar 4, 2025 by wwfu109 Loading…
1
Serialize using safetensors for KV caches ready ONLY add when PR is ready to merge/full CI is needed
#14228 opened Mar 4, 2025 by KuntaiDu Loading…
1
[V1][TPU] Support V1 Sampler for ragged attention ci/build v1
#14227 opened Mar 4, 2025 by NickLucche Loading…
1
[V1] Do not detokenize if sampling param detokenize is False v1
#14224 opened Mar 4, 2025 by hj-mistral Loading…
@njhill
4
[V1][Bugfix] Standardize quantized kv cache rejection for attention backends bug Something isn't working ready ONLY add when PR is ready to merge/full CI is needed v1
#14221 opened Mar 4, 2025 by mgoin Loading…
2
4
[V1][Metrics] Fix traceback with preemptions+LoRA v1
#14220 opened Mar 4, 2025 by markmc Loading…
3
[V1][PP] Support PP for MultiprocExecutor v1
#14219 opened Mar 4, 2025 by bigPYJ1151 Loading…
@comaniac
9
[Doc] Update nginx guide: remove privileged from vllm container run and add target GPU ID documentation Improvements or additions to documentation
#14217 opened Mar 4, 2025 by iacolippo Loading…
2
[Misc] Update compressed-tensors WNA16 to support zero-points
#14211 opened Mar 4, 2025 by dsikka • Draft
1
[Kernel] MoE tuning, quickly skip slow config
#14207 opened Mar 4, 2025 by noooop Loading…
1
[Model] Add Reasoning Parser for Granite Models documentation Improvements or additions to documentation frontend
#14202 opened Mar 4, 2025 by alex-jw-brooks Loading…
2
Moved numba from common requirements to cuda/rocm specific requirements ci/build force-merge ready ONLY add when PR is ready to merge/full CI is needed
#14199 opened Mar 4, 2025 by npanpaliya Loading…
2
docs: Add documentation for s390x cpu implementation documentation Improvements or additions to documentation
#14198 opened Mar 4, 2025 by dilipgb Loading…
1
[misc] Update blog link in README documentation Improvements or additions to documentation
#14194 opened Mar 4, 2025 by khluu Loading…
1
[RLHF] use worker_adapter_cls for compatibility with V0 and V1 ci/build documentation Improvements or additions to documentation
#14185 opened Mar 4, 2025 by youkaichao Loading…
1
[Bugfix] Correctly call cudaProfilerStop in benchmarks script
#14183 opened Mar 4, 2025 by b8zhong Loading…
1
Deepseek MTP for V1 needs-rebase v1
#14182 opened Mar 4, 2025 by sroy745 • Draft
2
Fix WorkerWrapperBase initialization: defer vllm_config setup
#14179 opened Mar 4, 2025 by vincent-4 Loading…
2
[DRAFT] Remove linear hack outside transformers backend
#14177 opened Mar 4, 2025 by Isotr0py • Draft
1
Add CUDA kernel for per_token_group_quant_fp8 ci/build
#14175 opened Mar 4, 2025 by mgoin • Draft
3
ProTip! Find all pull requests that aren't related to any open issues with -linked:issue.