Repository navigation

All pull requests

Labels (86)Milestones (1)New pull request

Open

(4,809)

Closed

(31,034)

[AMD] Multi-arch ROCm support for internal testing
#44308akao-amd opened Oct 14, 2026Contributor5 tasks
[Doc] cookbook: GB300 full-set accuracy and tool follow-up output for Ling-3.1-flash
#44307JustinTong0323 opened Oct 14, 2026Collaborator
1
[CI][Fix][NPU] Z-Image: Fix no synchronize after warmup request
#44306Allor-maker opened Oct 14, 2026Contributor2 of 5 tasks
[sgl-router] kv-indexer-bridge: receive event batches outside tokio's coop budget
#44305lixiufei-leo opened Oct 14, 2026Contributor3 tasks done
[Fix] glm5.2: Align HiSparse warmup backends and move test to nightly
#44303alphabetc1 opened Oct 14, 2026Collaborator3 of 4 tasks
4
[AMD][Qwen3.8]Enable fused GDN packing for ratio-3 heads
#44302amd-nprotaso opened Oct 14, 2026Contributor
[Feat] models: Ling-3.1-flash day-0 support, Hopper decode opts, prefill overlap, Ling3 fixes
#44300JustinTong0323 opened Oct 14, 2026Collaborator
2
[Feat] simulator: Add baseline collection and ML latency predictor training tools
#44299littlefatfat opened Oct 14, 2026Contributor5 tasks done
[Perf] [NVIDIA] DSA: shard k-pool prefill query rows across TP
#44298bvolpato opened Oct 14, 20264 of 5 tasks
[Perf] [AMD] kimi-k3: Fuse radix route with MXFP4 quant on the prequant path
#44297amd-yashagar opened Oct 14, 20264 of 5 tasks
[Fix] Reset global graph pool and allocator cache on graph cleanup and recapture
#44296devtyagi3909 opened Oct 14, 2026
[Fix] Require admin API key for flush_cache and LoRA endpoints
#44295bantianyue opened Oct 14, 2026
[Perf] [AMD] kimi-k3: Run shared-expert down GEMM through Aiter
#44294amd-yashagar opened Oct 14, 20263 of 5 tasks
support cp and bcg in GLM5.3-Flash
#44293yzk1989 opened Oct 14, 20265 tasks
[AMD] Enable torch.profiler GPU tracing on torch 2.12 / ROCm 10.1
#44292yctseng0211 opened Oct 14, 2026Collaborator5 tasks
[Bug] Fix peak concurrency accounting in bench_serving metrics
#44290saitarrun opened Oct 14, 2026
1
fix(speculative): add missing pp_proxy_tensors param to frozen_kv_mtp and uno spec-v2 workers
#44289Asthenia0412 opened Oct 14, 2026
1
[Perf] detokenizer: Group return IPC sends by tokenizer worker
#44288hope1262946533 opened Oct 14, 2026Contributor
[Feat] simulator: Support closed-loop agentic workloads
#44287Jeff010101 opened Oct 14, 20263 of 4 tasks
[Fix] diffusion: Fix VMoBA chunk selection bugs
#44286OnePunchMonk opened Oct 14, 2026Contributor
[PD] Add Mori Fabric backend for rack level gfx1250 serving
#44283inkcherry opened Oct 14, 2026Contributor5 tasks
[Perf] attention: Use nonzero(as_tuple=True) in GPU hot paths
#44282Asthenia0412 opened Oct 14, 2026
Add attention and indexer support for page-sharded DSA caches
#44280Shunkangz opened Oct 14, 2026Contributor5 tasks
[Fix] speculative: FR-Spec (--speculative-token-map) with a quantized (FP8) target lm_head
#44277AppNetDeveloper opened Oct 14, 20264 of 5 tasks
[Fix] benchmark: Count actual overlaps for peak concurrency
#44276znzryb opened Oct 14, 20263 tasks done

Pull requests