-
Notifications
You must be signed in to change notification settings - Fork 203
Pull requests: 1CatAI/1Cat-vLLM
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
fix(awq_sm70_moe): register int32 buffer Parameters with requires_grad=False
#630
opened Sep 14, 2026 by
jobordu
Loading…
[Core][SM70] Beat A800 Qwen3.8-27B FP8 C32 2K decode by 10%
#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…
[Cleanup][SM70] Drop the forced compile-cache opt-out for the Flash-V100 graph
#621
opened Sep 13, 2026 by
Peuqui
Loading…
[Bugfix][Distributed] Pass --distributed-timeout-seconds to the NCCL subgroups
#619
opened Sep 13, 2026 by
Peuqui
Loading…
[Bugfix][SM70/SM75] Gate the SM70 graph tunings on the worker's own pre-Ampere device
#618
opened Sep 13, 2026 by
Peuqui
Loading…
[Bugfix][Core] Backport upstream Mamba prefix-cache retention
#617
opened Sep 13, 2026 by
Leonccaa
Contributor
Loading…
[Bugfix][SM70] Fix decode collapse next to a long chunked prefill (#490)
#616
opened Sep 13, 2026 by
areslp
Loading…
[Bugfix][SM70/SM75] Honor a checkpoint's KV-cache quantization directive only on Ampere and newer
#613
opened Sep 12, 2026 by
Peuqui
Loading…
[Perf][SM70] Block-pack the activations of the NVFP4 QPN2 kernels
#611
opened Sep 12, 2026 by
Peuqui
Loading…
3 of 4 tasks
[Feature][SM75] Run ModelOpt NVFP4 and FP8 linears on Turing through the SM70 QPN kernels
#604
opened Sep 12, 2026 by
Peuqui
Loading…
3 of 4 tasks
[Bugfix][DeepSeek-V4] Align the SWA decode threshold with the sparse MLA builder
#603
opened Sep 12, 2026 by
Peuqui
Loading…
3 of 4 tasks
[Bugfix][Build][SM70] Declare the pybind11 SM70 extensions non-limited-API
#601
opened Sep 11, 2026 by
Peuqui
Loading…
3 of 4 tasks
[Bugfix][Platform] Resolve an unspecified device_id to the worker's own device
#600
opened Sep 11, 2026 by
Peuqui
Loading…
3 of 4 tasks
[Bugfix][Spec Decode][SM70] Gate DFlash2's BF16 emulation and FlashInfer top-k on the worker's own device
#599
opened Sep 11, 2026 by
Peuqui
Loading…
3 of 4 tasks
[KV Offload] Grouped hybrid RAM caching, filesystem restoration and retention-sized Mamba pools (stacked on #617)
#598
opened Sep 11, 2026 by
Leonccaa
Contributor
Loading…
[Perf] Reuse verified H3 prepared weights across model loads
#593
opened Sep 10, 2026 by
yangzhuxinyzx
Contributor
•
Draft
[Bugfix] DFlash: fuse context K/V through quant_method so a quantized draft head loads
#592
opened Sep 10, 2026 by
Peuqui
Loading…
3 of 4 tasks
Bump the minor-update group across 1 directory with 139 updates
dependencies
Pull requests that update a dependency file
#590
opened Sep 10, 2026 by
dependabot
Bot
Loading…
[Video] Add native H3 request-scoped TeaCache and Cache-DiT
#584
opened Sep 9, 2026 by
yangzhuxinyzx
Contributor
•
Draft
[Kernel] Share SM70 projection, attention and row reduction paths
#581
opened Sep 9, 2026 by
yangzhuxinyzx
Contributor
•
Draft
[Video] Measure H3 workflows with actual denoise schedules
#578
opened Sep 9, 2026 by
yangzhuxinyzx
Contributor
•
Draft
Previous Next
ProTip!
Add no:assignee to see everything that’s not assigned.