Skip to content

Pull requests: InfiniTensor/InfiniLM

New pull request New
87 Open 302 Closed
87 Open 302 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(nvidia): add Mamba2 and RWKV5 adapters
#591 opened Sep 20, 2026 by lgeln10 Loading…
7 of 8 tasks
perf: reduce KV admission over-reservation
#590 opened Sep 20, 2026 by lgeln10 Loading…
feat(nvidia): 为 v0.2.9c 增加可复用的 GGUF Route B 支持
#589 opened Sep 20, 2026 by xindongliu594 Loading…
40 of 46 tasks
feat: add mtp draft models for speculative decoding
#588 opened Sep 20, 2026 by accelerator-llc Loading…
34 of 41 tasks
feat(model): support LFM2-1.2B
#587 opened Sep 20, 2026 by Hyacinth25 Loading…
11 of 16 tasks
1
feat: add FP8 blockwise weight quantization and FP8(E4M3) KV cache support
#586 opened Sep 20, 2026 by shsaihdsaiudh Loading…
38 of 42 tasks
perf: Qwen3 offline inference — hybrid attention backend, fused rms_norm_rope, chunked prefill + prompt-lookup spec decoding
#585 opened Sep 20, 2026 by shsaihdsaiudh Loading…
feat(qwen): add paged greedy MTP with FP8 block weights
#584 opened Sep 19, 2026 by big-hip Loading…
40 of 41 tasks
feat(minimax): support MiniMax-Text-01 and MiniMax-M2
#583 opened Sep 19, 2026 by y258dd Loading…
25 of 42 tasks
feat(quantization): support compressed-tensors W8A8 and mixed modules
#582 opened Sep 19, 2026 by sam0336 Loading…
35 of 44 tasks
1
Feat/add gemma3 model
#581 opened Sep 18, 2026 by zyc107109102 Loading…
feat(nvidia): support model type: granitemoehybrid
#580 opened Sep 18, 2026 by WHoutstanding Loading…
41 of 49 tasks
perf: reuse running reservations within each scheduler step
#579 opened Sep 17, 2026 by T4t4KAU Loading…
9 of 11 tasks
perf: bound KV page reclamation candidate scans
#578 opened Sep 17, 2026 by T4t4KAU Loading…
8 of 10 tasks
perf: build KV slot mappings by page ranges
#577 opened Sep 17, 2026 by T4t4KAU Loading…
9 of 11 tasks
perf: avoid scanning KV cache pages for capacity checks
#576 opened Sep 17, 2026 by T4t4KAU Loading…
15 of 22 tasks
feat(mamba2): integrate model loading and recurrent inference
#575 opened Sep 17, 2026 by big-hip Loading…
38 of 39 tasks
Feat/OpenAI agent compat
#574 opened Sep 17, 2026 by wooway777 Collaborator • Draft
49 tasks
feat(engine): integrate prefix eviction and chunked parallel execution
#573 opened Sep 16, 2026 by big-hip Loading…
32 of 41 tasks
1
feat(nvidia): support MiniMax-Text-01
#572 opened Sep 15, 2026 by Kritace Loading…
29 of 34 tasks
feat: add priority-aware request scheduling
#571 opened Sep 13, 2026 by xiaoba17 Loading…
26 of 36 tasks
1
fix(moore): enable Mate flash-attn inference
#567 opened Sep 10, 2026 by spike-zhu Collaborator Loading…
1 of 49 tasks
feat(nvidia): add reusable GGUF Route B support for Qwen3.5
#559 opened Sep 3, 2026 by xindongliu594 Loading…
32 of 48 tasks
1
perf: fuse unquantized Qwen3-Next GDN projections
#555 opened Sep 2, 2026 by T4t4KAU Loading…
26 of 39 tasks
feat(server): add agent support with tool-call and reasoning parsing
#554 opened Sep 1, 2026 by rubik-hua Contributor Loading…
7
ProTip! no:milestone will show everything without a milestone.