Skip to content

Pull requests: vllm-project/speculators

New pull request New
25 Open 248 Closed
25 Open 248 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
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Fix small typos in ONLINE_TRAINING.md documentation Improvements or additions to documentation
#371 opened Mar 27, 2026 by dsikka Loading…
2
fix: resolve runtime crashes and mypy errors with transformers >= 4.57 + Pydantic v2
#369 opened Mar 27, 2026 by rahul-tuli • Draft
1
6
feat(eval): add MTP built-in head support to eval-guidellm framework documentation Improvements or additions to documentation quality-failed
#367 opened Mar 26, 2026 by rahul-tuli Loading…
@rahul-tuli
5
feat(fast-mtp): FastMTP speculative head finetuning for Qwen3-Next documentation Improvements or additions to documentation
#366 opened Mar 26, 2026 by rahul-tuli • Draft
1
fix(datagen): VllmHiddenStatesGenerator fails for hybrid-attention models and stalls on large datasets
#365 opened Mar 26, 2026 by rahul-tuli • Draft
1
[Doc] Add a choosing algorithms section documentation Improvements or additions to documentation
#355 opened Mar 22, 2026 by DonaghBr Loading…
4 tasks done
1
[Feat]Add DFlash training support quality-failed
#354 opened Mar 20, 2026 by shanjiaz Loading…
4 tasks
33
PR4: FastMTP training support — dataset labels, trainer utils, and frozen embeddings documentation Improvements or additions to documentation
#349 opened Mar 17, 2026 by rahul-tuli • Draft
3 tasks
1
[WIP]update generator to support vllm latest main
#348 opened Mar 16, 2026 by shanjiaz • Draft
3 of 4 tasks
2
add multimodal preprocessing support
#344 opened Mar 13, 2026 by shx2005 Loading…
3 of 4 tasks
6
[WIP] phase 1 of peagle support needs-rebase
#343 opened Mar 13, 2026 by shanjiaz • Draft
2 of 4 tasks
3
Add FastMTP data generation and dataset utilities documentation Improvements or additions to documentation
#342 opened Mar 12, 2026 by rahul-tuli • Draft
1
Add FastMTP checkpoint converter documentation Improvements or additions to documentation
#340 opened Mar 12, 2026 by rahul-tuli • Draft
1
2
add speculators finetune examples documentation Improvements or additions to documentation needs-rebase
#339 opened Mar 10, 2026 by Annarine Loading…
4 tasks
3
feat: P-EAGLE Phase 1 - COD sampling, model, config, and loss utilities needs-rebase
#327 opened Mar 4, 2026 by NJX-njx Loading…
1
11
docs: add algorithm documentation for EAGLE-3 and FastMTP documentation Improvements or additions to documentation
#326 opened Mar 4, 2026 by NJX-njx Loading…
7
Avoid D2H copies in VllmHiddenStatesGenerator to speed up offline data generation by x5
#322 opened Mar 2, 2026 by qGentry Loading…
5
Materialize non-layer modules after FSDP sharding
#314 opened Feb 27, 2026 by guan404ming Loading…
7
Add FastMTP multi-token prediction speculator documentation Improvements or additions to documentation
#301 opened Feb 24, 2026 by rahul-tuli • Draft
25
Feature/tool use support
#300 opened Feb 24, 2026 by olisicky Loading…
10
add multinode train support
#299 opened Feb 24, 2026 by Liccol Loading…
14
support sp ulysses needs-rebase
#298 opened Feb 24, 2026 by momo609 Loading…
10
Feat: FastMtp Model Implementation Substack 2
#281 opened Feb 8, 2026 by ved1beta Loading…
1
14
Add Eagle3 training smoke tests
#201 opened Dec 3, 2025 by fynnsu • Draft
1 of 4 tasks
3
Initial MLP Speculative Decoding implementation
#160 opened Oct 28, 2025 by christinaexyou Loading…
5
ProTip! Find all pull requests that aren't related to any open issues with -linked:issue.