Skip to content

Pull requests: HabanaAI/vllm-hpu-extension

New pull request New
24 Open 218 Closed
24 Open 218 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

Allow usage of fused_block_softmax_adjustment for Qwen with Lazy
#246 opened Jun 27, 2025 by mswiniarsk • Draft
Switch softmax mode to fast for DeepSeek-R1 inference
#244 opened Jun 27, 2025 by Wei-Lin-Intel Loading…
1
enable fp32 softmax for flat_pa_mla
#243 opened Jun 27, 2025 by yangulei Loading…
Update dependabot.yml
#242 opened Jun 26, 2025 by michalkuligowski Loading…
Update linear.py
#239 opened Jun 25, 2025 by michalkuligowski Loading…
Integrating block_softmax
#238 opened Jun 24, 2025 by ksmusz • Draft
Remove double generate
#229 opened Jun 18, 2025 by adobrzyn Loading…
[DeepSeek] Add automatic Pile-10k dataset processing and extended calibration settings ready for review
#225 opened Jun 16, 2025 by yiliu30 Loading…
13
Exponential bucketing tweaks
#224 opened Jun 13, 2025 by madamczyk-intel Loading…
Bucketing refactoring
#223 opened Jun 12, 2025 by adobrzyn Loading…
14
[WIP] Add runtime conversion of fp8fn models to fp8fnuz
#222 opened Jun 12, 2025 by kwisniewski98 Loading…
2
Find bucket with bmin not divs by step
#212 opened Jun 5, 2025 by adobrzyn Loading…
1
Fix max_blocks for warmup decode buckets in case of disabled CONTIGUOUS PA feature
#204 opened May 29, 2025 by iboiko-habana Loading…
1
Use sets for faster filter checks. Better long context support
#203 opened May 28, 2025 by pi314ever Loading…
3
Add useful internal vllm test
#200 opened May 27, 2025 by nirda7 • Draft
4
[SW-225565] Enable triangular softmax with merged prefill
#197 opened May 26, 2025 by kamil-kaczor • Draft
2
fix the issue that bmax not in bucket buffer
#191 opened May 22, 2025 by sywangyi Loading…
4
Unify FusedMoe with expert parallelism
#175 opened May 14, 2025 by mengniwang95 Loading…
17
Optimized MoE on Gaudi
#159 opened Apr 18, 2025 by gyou2021 • Draft
36
[FIX] fp8 gc compile error
#110 opened Mar 4, 2025 by maktukmak • Draft
1
Expand capability checks
#89 opened Feb 3, 2025 by kzawora-intel • Draft
9
Add renormalize parameter for FusedMOE's & modify experts_max arg of mixture_of_experts()
#70 opened Jan 9, 2025 by tangleintel • Draft
4
[WIP] Add option to do group sum on TPC instead of MME
#64 opened Dec 20, 2024 by mswiniarsk • Draft
Remove vllm.logger.init_logger dependency
#53 opened Dec 9, 2024 by kzawora-intel • Draft
2
ProTip! Follow long discussions with comments:>50.