Skip to content

Pull requests: EfficientMoE/MoE-Infinity

New pull request New
22 Open 124 Closed
22 Open 124 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(dispatcher): stream-order expert exec against input producer and consumers
#197 opened Sep 3, 2026 by drunkcoding Contributor Loading…
feat(serving): DeepSeek-V2/V3 paged-attention shim (MLA) — fixes #191
#195 opened Sep 3, 2026 by drunkcoding Contributor Loading…
1
feat(#184): adaptive-precision benchmark harnesses + release-gate evaluation
#194 opened Sep 2, 2026 by drunkcoding Contributor Loading…
fix(#178): complete chunked prefill — KV budget cap, FlashInfer workspace, ragged attention
#193 opened Sep 2, 2026 by drunkcoding Contributor Loading…
fix(serving): fail loud when a served model has no paged-attention shim
#192 opened Sep 2, 2026 by drunkcoding Contributor Loading…
fix(serving): prefix-cache binding, hit stats, dtype (needs #197)
#190 opened Aug 31, 2026 by drunkcoding Contributor • Draft
fix(serving): serialize SSE generator + FlashInfer plan dtype (bf16)
#189 opened Aug 30, 2026 by drunkcoding Contributor Loading…
1
fix(serving): proactive decode-headroom preemption to trigger KV swap
#188 opened Aug 30, 2026 by drunkcoding Contributor Loading…
fix: single-owner kTopologyHandle guard + adaptive-memory benchmark hardening
#187 opened Aug 30, 2026 by drunkcoding Contributor Loading…
fix(kernel): fused_decode_attention raises TypeError whenever MOE_DISABLE_FUSED_KERNELS=1
#186 opened Aug 30, 2026 by Anai-Guo Loading…
6 of 9 tasks
1
feat: overlap-aware expert prefetch
#180 opened Aug 21, 2026 by drunkcoding Contributor Loading…
6 tasks done
2
feat: native-paged decode CUDA graphs
#182 opened Aug 21, 2026 by drunkcoding Contributor Loading…
7 tasks done
2
feat: GPU-only expert routing (native CUDA active-expert discovery + stream-ordered completion)
#176 opened Aug 21, 2026 by drunkcoding Contributor Loading…
7 tasks done
2
feat: adaptive expert/KV memory allocation
#177 opened Aug 21, 2026 by drunkcoding Contributor Loading…
10 of 11 tasks
2
feat: adaptive mixed-precision expert cache
#184 opened Aug 21, 2026 by drunkcoding Contributor Loading…
8 tasks done
6
feat: asynchronous hierarchical KV swap
#185 opened Aug 21, 2026 by drunkcoding Contributor Loading…
9 tasks done
2
feat: opt-in INT8 KV-cache quantization
#183 opened Aug 21, 2026 by drunkcoding Contributor Loading…
6 tasks done
2
feat: prefix KV reuse
#181 opened Aug 21, 2026 by drunkcoding Contributor Loading…
6 tasks done
3
feat: chunked prefill scheduling (bounded, decode-first, opt-in)
#178 opened Aug 21, 2026 by drunkcoding Contributor Loading…
5 tasks done
3
feat: phase-specific expert policy
#179 opened Aug 21, 2026 by drunkcoding Contributor Loading…
8 tasks done
3
Promote dev → main: unified DFlash execution and PyPI versioning
#175 opened Aug 21, 2026 by drunkcoding Contributor • Draft
Fix OLMoE support and silent model-init failures
#119 opened Jul 18, 2026 by danielesalpietro Contributor Loading…
6 tasks
2
ProTip! Follow long discussions with comments:>50.