-
Notifications
You must be signed in to change notification settings - Fork 183
Pull requests: alibaba/rtp-llm
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
fix(rocm): apply RoPE for embedding models without KV cache
#973
opened May 7, 2026 by
siluzhou
Collaborator
Loading…
async schedule [3/N]: support async scheduler
#972
opened May 7, 2026 by
Vinkle-hzt
Collaborator
Loading…
fix(rocm): propagate hw_kernel_config to qwen35 layers
#970
opened May 7, 2026 by
chengshu-lcc
Collaborator
Loading…
feat: enable Qwen35-MoE MTP && sp prefill cuda graph
#969
opened May 7, 2026 by
amd-yilizhao
Collaborator
Loading…
feat: PureCP/PureDP allgather+RS routing for FP8 per-block MoE
#968
opened May 6, 2026 by
intermezzi
Collaborator
Loading…
feat: add DeepGEMM JIT kernel warmup integrated into C++ engine startup
#967
opened May 6, 2026 by
ydshi0
Loading…
[WIP] feat: add dp controller master support (current only rr strategy)
#963
opened May 2, 2026 by
bppps
Collaborator
Loading…
feat(deps): unify pip deps via PEP 503 indexes + thin requirements
#962
opened Apr 30, 2026 by
LLLLKKKK
Collaborator
Loading…
1 of 2 tasks
feat: implement CpuTpBroadcaster for CPU-only tensor broadcasting
#960
opened Apr 30, 2026 by
Vinkle-hzt
Collaborator
Loading…
fix - make prepare_cg_spec_decode_kernel easy use and understand
#954
opened Apr 29, 2026 by
zerozw
Collaborator
Loading…
feat: suport hybrid pool kvcache allocator
#943
opened Apr 28, 2026 by
SJTUGavinLiu
Collaborator
Loading…
async schedule [2/N]: support async prepare
#936
opened Apr 26, 2026 by
Vinkle-hzt
Collaborator
Loading…
fix: fix rocm greedy sampling to avoid crash
#932
opened Apr 24, 2026 by
liaocz
Collaborator
Loading…
feat(rocm): MoRI EP (Expert Parallelism) support for MI355X
#931
opened Apr 24, 2026 by
jacobwin-ai
Collaborator
Loading…
[fix] Handle enqueue failures in RPC and API paths
#929
opened Apr 23, 2026 by
ZhihanYan
Collaborator
Loading…
Previous Next
ProTip!
no:milestone will show everything without a milestone.