Skip to content

Pull requests: alibaba/rtp-llm

New pull request New
111 Open 871 Closed
111 Open 871 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

feat: add RDMA transport for ViT
#1105 opened Jun 15, 2026 by ydshi0 Loading…
1
feat(cache): migrate DSV4 KV cache infrastructure from feat/dsv4_on_d…
#1103 opened Jun 15, 2026 by Adrenaline-S Collaborator • Draft
[ROCm] Prefill performance optimization for embedding models
#1102 opened Jun 15, 2026 by liaocz Collaborator Loading…
3
Feature/vit rpc metrics worker status
#1101 opened Jun 15, 2026 by yzyDavid Collaborator Loading…
2
Feat/support dash frontend
#1100 opened Jun 15, 2026 by wanglining97 Collaborator Loading…
2
fix: size flashinfer prefill workspace dynamically
#1099 opened Jun 15, 2026 by Vinkle-hzt Collaborator Loading…
3
fix(moe): avoid cross-warp stale read in ep_scatter_1 prefix sum
#1098 opened Jun 15, 2026 by HoniiTro19 Loading…
2
generic layer-wise KV cache specs
#1097 opened Jun 15, 2026 by Adrenaline-S Collaborator Loading…
2
feat(xpu): XPU module factories and bindings (4/4)
#1096 opened Jun 13, 2026 by aslanxie Loading…
8
feat(online_optimizer) support theoretical hit rate statistics in flexlb
#1095 opened Jun 12, 2026 by YoungRX Collaborator Loading…
3
Feat/adapt sm120 rtx5000pro merge
#1094 opened Jun 12, 2026 by parkerpang Collaborator Loading…
4
Feature/offline inference scheduling
#1093 opened Jun 11, 2026 by HoniiTro19 Loading…
4
perf: remove redundent kv cache update for finished stream
#1092 opened Jun 11, 2026 by zhangjianning-zjn Collaborator Loading…
4
fix: fix 1s stall during polling output of generate stream
#1091 opened Jun 11, 2026 by zhangjianning-zjn Collaborator Loading…
1
fix: fix metric reporting on waiting time of generate stream
#1090 opened Jun 11, 2026 by zhangjianning-zjn Collaborator Loading…
4
perf(rocm): enable FlyDSL fused MoE for MI308X Qwen3.5 decode
#1087 opened Jun 11, 2026 by chengshu-lcc Collaborator Loading…
1
feat(rocm): support Qwen3/3.5 VL model on ROCm
#1086 opened Jun 11, 2026 by liaocz Collaborator Loading…
10
feat(moe): fuse shared expert into MoE kernel for ROCm TP-only mode
#1085 opened Jun 11, 2026 by chengshu-lcc Collaborator Loading…
14
feat: add prompt scoring (per-position logits for input sequences)
#1081 opened Jun 10, 2026 by theNiemand Collaborator Loading…
7
feat: add simple scheduler for ViT
#1079 opened Jun 9, 2026 by ydshi0 Loading…
16
feat(xpu): Python device abstraction and server integration (3/4)
#1075 opened Jun 8, 2026 by aslanxie Loading…
14
feat(omni): add Qwen2.5-Omni multi-stage pipeline support
#1074 opened Jun 7, 2026 by stmatengss Loading…
1 of 3 tasks
8
feat(new_weight_loader): introduce new weight loader framework with FP8 quantization support for Qwen2
#1072 opened Jun 5, 2026 by Oneydauh Loading…
5
feat(xpu): C++ device generalization for Intel XPU support (2/4)
#1071 opened Jun 5, 2026 by aslanxie Loading…
41
feat(p2p): decode_entrance P2P support with lease-based race fix
#1067 opened Jun 4, 2026 by ZhihanYan Collaborator Loading…
7 tasks
27
ProTip! Updated in the last three days: updated:>2026-06-12.