-
Notifications
You must be signed in to change notification settings - Fork 218
Pull requests: 1CatAI/1Cat-vLLM
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[Bugfix][KV Offload] Order SimpleCPUOffloadConnector stores behind the compute stream
#665
opened Sep 20, 2026 by
Peuqui
Loading…
[Model][SM70] Calibrated E4M3 KV cache for Qwen4Exp QSA with MTP
#664
opened Sep 20, 2026 by
Leonccaa
Contributor
Loading…
[Bugfix][CI/Build][SM70] Copy H3 source trees into Docker csrc build
#663
opened Sep 20, 2026 by
carrey-feng
Contributor
Loading…
4 tasks done
[Bugfix][Spec Decode] Ship the speculative round state to non-last PP ranks
#662
opened Sep 19, 2026 by
Peuqui
Loading…
3 of 4 tasks
[Bugfix][SM70] Pin decode workspaces allocated during CUDA graph capture
#661
opened Sep 19, 2026 by
Leonccaa
Contributor
Loading…
[Bugfix][SM70] Release Flash-V100 prefill workspaces before growing them
#660
opened Sep 19, 2026 by
Leonccaa
Contributor
Loading…
[Feature][Spec Decode] Apply min_p in the rejection sampler
#659
opened Sep 19, 2026 by
Peuqui
Loading…
3 of 4 tasks
[Bugfix][DeepSeek-V4] Keep mHC finite under float16 for attention-sink rows
#658
opened Sep 19, 2026 by
Peuqui
Loading…
3 of 4 tasks
[Bugfix][Core] Backport vllm-project/vllm#44082: cache the EAGLE/MTP lookahead block in the SWA prefix-cache mask
#657
opened Sep 19, 2026 by
Peuqui
Loading…
3 of 4 tasks
feat: add SM72 (Jetson AGX Xavier) compute capability support
#655
opened Sep 19, 2026 by
hernandez42
Loading…
[Bugfix] PLE offload: send registration inputs by file descriptor
#654
opened Sep 18, 2026 by
Peuqui
Loading…
3 of 4 tasks
[Bugfix] Constrain tools + response_format with a union grammar in Qwen3XMLToolParser
#653
opened Sep 18, 2026 by
areslp
Contributor
Loading…
[Model][SM70] Accept DFlash embed replica across split load_weights c…
#651
opened Sep 17, 2026 by
ProprietaryLegal
Loading…
4 tasks
[Bugfix][SM70] Stop grouped-page4 null-block padding from leaking E4M3 NaN
#647
opened Sep 17, 2026 by
areslp
Contributor
Loading…
[Feature][Qwen4Exp] PLE overflow cascade: device, pinned host, spare GPU, disk
#646
opened Sep 16, 2026 by
Peuqui
Loading…
4 tasks done
[Bugfix][Qwen4Exp] Detect the FP8 PLE table from ModelOpt mixed-precision configs
#640
opened Sep 15, 2026 by
Peuqui
Loading…
3 of 4 tasks
[Bugfix][Qwen4Exp] Allow checkpoint FP8 MTP experts under pipeline parallelism
#639
opened Sep 15, 2026 by
Peuqui
Loading…
4 tasks done
[Bugfix][MRV2] Backport phase-aware decode graph dispatch
#637
opened Sep 14, 2026 by
Leonccaa
Contributor
Loading…
[Bugfix][Qwen3.5] Keep the MTP drafter stage-local under pipeline parallelism
#636
opened Sep 14, 2026 by
Peuqui
Loading…
3 of 4 tasks
fix(awq_sm70_moe): register int32 buffer Parameters with requires_grad=False
#630
opened Sep 14, 2026 by
jobordu
Loading…
[Core][SM70] Beat A800 C32 decode and recover 32K serving
#629
opened Sep 14, 2026 by
yangzhuxinyzx
Contributor
•
Draft
[Bugfix][Spec Decode] Reject out-of-range sampled token ids in the async output path
#625
opened Sep 13, 2026 by
publee
Loading…
[CI] Fix baseline QSA typing and pre-commit failures
#624
opened Sep 13, 2026 by
Leonccaa
Contributor
Loading…
[Feature][SM75] Build and load a Turing FlashAttention-2 library next to the Volta one
#623
opened Sep 13, 2026 by
Peuqui
Loading…
[Bugfix][Qwen4Exp] Resolve the PLE table pointer inside the gather op instead of baking it into the graph
#622
opened Sep 13, 2026 by
Peuqui
Loading…
Previous Next
ProTip!
Adding no:label will show everything without a label.