Search results

[Feature] HiCache Support for SWA (Gemma and Mistral)

Status: Open.

#23659 In sgl-project/sglang;

· baoskee opened on Apr 24, 2026

[Bug] DeepSeek-V4 compressed attention backend: no SM120 fallback for Lightning Indexer

Status: Open.

#23657 In sgl-project/sglang;

· rs-ipps opened on Apr 24, 2026

[Bug] [NPU] Qwen3.5-35B-A3B missing 1 required positional argument: 'interpolation'

Status: Open.

#23656 In sgl-project/sglang;

· OrangeRedeng opened on Apr 24, 2026

[Feature] [NPU] not supported prefix cache for Qwen3.5

Status: Open.

#23652 In sgl-project/sglang;

· LilyPankratova opened on Apr 24, 2026

[Question][Diffusion][cache] How does the low-step diffusion models benefit from cache?

Status: Open.

#23640 In sgl-project/sglang;

· JohnHerry opened on Apr 24, 2026

[Feature] UnifiedRadix support hicache for DeepSeekV4

Status: Open.

#23639 In sgl-project/sglang;

· hzh0425 opened on Apr 24, 2026

[Bug] DP scheduler workers escape Slurm's job cgroup and survive job termination, blocking GPU reuse

Status: Open.

#23627 In sgl-project/sglang;

· DavidBellamy opened on Apr 24, 2026

[Bug] silu_and_mul and other activation kernels crash with cudaErrorInvalidConfiguration when num_tokens=0

Status: Open.

#23609 In sgl-project/sglang;

· imReese opened on Apr 24, 2026

[Bug] Multi-node deployment crashes with collective sequence number mismatch in scheduler initialization (Gemma4)

Status: Open.

#23603 In sgl-project/sglang;

· solitude-alive opened on Apr 24, 2026

DeepSeek V4 Roadmap

deepseek

high priority

roadmap

Status: Open.

#23602 In sgl-project/sglang;

· fzyzcjy opened on Apr 24, 2026

DeepSeek-V4 Day 0 Support on NPUs

Status: Open.

#23598 In sgl-project/sglang;

· iforgetmyname opened on Apr 24, 2026

[Bug] Hy3-preview cuda-graph crash on AMD MI300X/MI355X due to AITER custom all-reduce stream invalidation

Status: Open.

#23580 In sgl-project/sglang;

· andyluo7 opened on Apr 23, 2026