Skip to content

Pull requests: RBLN-SW/vllm-rbln

New pull request New
32 Open 548 Closed
32 Open 548 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(attn): normalize seq_lens to int32 so custom kernel compiles torch.compile torch.compile based implementation
#648 opened Jun 5, 2026 by rebel-jinhwan Contributor Loading…
13 tasks
@rebel-jinhwan
1
other(torch_compile): parametrize e2e test over RBLN_WEIGHT_FREE (wf_on xfail) torch.compile torch.compile based implementation
#647 opened Jun 5, 2026 by rebel-jinhwan Contributor Loading…
1 of 13 tasks
@rebel-jinhwan
1
feature(warmup): offload host regions during device_tensor warm-up
#646 opened Jun 5, 2026 by rebel-jonghewk Collaborator • Draft
13 tasks
1
feature(env): default VLLM_RBLN_AUTO_PORT to VLLM_RBLN_USE_DEVICE_TENSOR
#645 opened Jun 4, 2026 by rebel-jonghewk Collaborator Loading…
1 of 13 tasks
5
fix: apply classifier activation in RBLNClassifierPooler
#644 opened Jun 4, 2026 by rebel-thkim Contributor Loading…
1
fix: pass hf_config to optimum compilation
#643 opened Jun 4, 2026 by rebel-thkim Contributor Loading…
1
model: assert when executing an unregistered model
#642 opened Jun 4, 2026 by rebel-eunji Collaborator Loading…
1 of 13 tasks
@rebel-eunji
1
feature(nixl): support nixl-rbln
#640 opened Jun 2, 2026 by rebel-yskim Contributor • Draft
13 tasks
@rebel-yskim
1
model: Gemma4 31B E2B and E4B torch.compile torch.compile based implementation
#632 opened May 29, 2026 by pei0033 Collaborator Loading…
4 of 13 tasks
@kkimmk @pei0033
1
feature: implement vllm benchmark tracing for analysis
#628 opened May 28, 2026 by rebel-wonsubkim Contributor Loading…
1 of 13 tasks
@rebel-wonsubkim
4
model: support deepseek v3
#624 opened May 26, 2026 by rebel-kblee Contributor • Draft
1 of 13 tasks
1
[DO NOT MERGE] model: support gemma4
#622 opened May 26, 2026 by rebel-eunji Collaborator • Draft
13 tasks
@rebel-eunji
1
feature: introduce mega cache
#619 opened May 22, 2026 by rebel-jonghewk Collaborator Loading…
3 tasks done
1
other: switch vllm source to rebellions internal pypi
#614 opened May 19, 2026 by rebel-minhopark Loading…
3 of 13 tasks
2
feature: enable all2all communication for MoE model
#605 opened May 14, 2026 by rebel-jangys Contributor Loading…
13 tasks
15
other: bump to match upstream v0.19.1 (optimum)
#603 opened May 13, 2026 by rebel-eunji Collaborator Loading…
1 of 13 tasks
@rebel-eunji
2
feat: enable request-reordering in optimum-compiled models
#596 opened May 11, 2026 by rebel-thkim Contributor • Draft
13 tasks
2
other: improve optimum-compile
#591 opened May 7, 2026 by rebel-seinpark Collaborator Loading…
13 tasks
3
refactor: ci visibility improvement
#584 opened May 4, 2026 by rebel-seinpark Collaborator • Draft
13 tasks
1
feature(runtime): async exec + async memcpy
#581 opened Apr 29, 2026 by rebel-jonghewk Collaborator Loading…
3 of 5 tasks
6
other(bumpup): Make vllm-rbln type2 compatible to upstream v0.19.1 torch.compile torch.compile based implementation
#544 opened Apr 20, 2026 by rebel-jinhwan Contributor Loading…
1 of 13 tasks
@rebel-jinhwan
5
feature: support for rejection sampler (WIP)
#543 opened Apr 19, 2026 by rebel-eunji Collaborator • Draft
13 tasks
@rebel-eunji
1
model: ax k1 fp8
#541 opened Apr 17, 2026 by rebel-thkim Contributor Loading…
5 of 13 tasks
2
feature: pd disagregation docker draft
#510 opened Apr 7, 2026 by rebel-jindol21 Contributor Loading…
13 tasks
feature(metrics): dump stat to a file
#480 opened Mar 25, 2026 by rebel-minwook Loading…
4 tasks
1
ProTip! Type g p on any issue or pull request to go back to the pull request listing page.