All issues

Search results

Does SGLang have a demo for running the VBench dataset accuracy evaluation on Wan2.2?

Status: Open.

#37238 In sgl-project/sglang;

· maidouxiaozi opened on Aug 31, 2026

[Bug] remote_instance weight loading fails with --pp-size 2 — TransferEngine info keyed by tp_rank

Status: Open.

#37216 In sgl-project/sglang;

· twb1235 opened on Aug 31, 2026

[Bug] --dp 8 intermittently fails with TCPStore EADDRINUSE on single-node 8×H800

Status: Open.

#37215 In sgl-project/sglang;

· Ben3892 opened on Aug 31, 2026

[Bug] GPT-OSS with DP attention enabled produces garbage output (GSM8K 0/128 vs 124/128)

Status: Open.

#37187 In sgl-project/sglang;

· sghhhh opened on Aug 31, 2026

[Bug] Qwen3.5 visual encoder weights dropped by PP layer-range filter in load_weights

Status: Open.

#37186 In sgl-project/sglang;

· BakerBunker opened on Aug 31, 2026

AMD MI308X SGLang GLM-5.3-Flash ValueError: The checkpoint you are trying to load has model type glm5_next but Transformers does not recognize this architecture.

Status: Open.

#37183 In sgl-project/sglang;

· Alan-D-Chen opened on Aug 31, 2026

LMCache integration ignores per-request cache_salt: external storage keys are token-only, breaking tenant isolation

Status: Open.

#37160 In sgl-project/sglang;

· edurdias opened on Aug 30, 2026

[Feature] Tuning / per-GPU config for DSv4 top-k v2 cluster launch plan (kClusterFloor / kNumPersistentClusters / kCandidates)

Status: Open.

#37150 In sgl-project/sglang;

· dahaobenhao opened on Aug 30, 2026

[Bug] Spec V2 paths no longer emit speculative-decoding OpenTelemetry spans

Status: Open.

#37128 In sgl-project/sglang;

· RichardoMrMu opened on Aug 30, 2026

[Bug] Qwen3.8-Flash-Next QSA + NEXTN decode graph silently corrupts output on GB10 TP2

Status: Open.

#37111 In sgl-project/sglang;

· hellojiaru opened on Aug 30, 2026

[Bug] GLM-5.3-Flash on RTX PRO 6000 (sm_120): two DSA backend blockers after the deep_gemm NameError

Status: Open.

#37105 In sgl-project/sglang;

· Hansen523 opened on Aug 30, 2026

[Bug] Pretokenized image requests can crash GLM MRoPE with a stale retokenized mask

Status: Open.

#37097 In sgl-project/sglang;

· nanjiangwill opened on Aug 30, 2026