Search results

[Bug] Intermittent segfault in Triton MoE kernel during piecewise CUDA graph warmup on B200

Status: Open.

#21629 In sgl-project/sglang;

· yhyang201 opened on Mar 29, 2026

Feature: Add TurboQuant KV Cache Quantization for Memory-Efficient Long-Context Inference

Status: Open.

#21618 In sgl-project/sglang;

· scottgl9 opened on Mar 29, 2026

[Bug] Qwen3.5-397B-A17B-FP8 produces gibberish at batch concurrency >= 32 on AMD MI355X

Status: Open.

#21614 In sgl-project/sglang;

· Arist12 opened on Mar 29, 2026

[Bug] mistral small 4 fails to start

Status: Open.

#21611 In sgl-project/sglang;

· xhejtman opened on Mar 28, 2026

[Bug] ModelOptModelLoader doesn't support --load-format sharded_state (NVFP4 models)

Status: Open.

#21603 In sgl-project/sglang;

· vroomfondel opened on Mar 28, 2026

[Bug] modelopt_quant.py: NVFP4 input_scale not sliced to local experts with EP > 1

Status: Open.

#21602 In sgl-project/sglang;

· vroomfondel opened on Mar 28, 2026

[RFC][NPU] Ascend NPU A5 Support for MXFP8/MXFP4 Quantization

Status: Open.

#21584 In sgl-project/sglang;

· YChange01 opened on Mar 28, 2026

[Bug] [ROCM] Qwen3.5-397B-A17B-FP8 Generates Garbage Response

Status: Open.

#21566 In sgl-project/sglang;

· zyzshishui opened on Mar 27, 2026

[Feature] [Diffusion] Will sglang-diffusion support LTX2.3-22b?

Status: Open.

#21557 In sgl-project/sglang;

· vibe-viscot opened on Mar 27, 2026

[Feature] Implement defunctionalization for sgl_kernel ops in FixFunctionalizationPass

Status: Open.

#21550 In sgl-project/sglang;

· Archie-wang opened on Mar 27, 2026

[MPS] Triton stub sub-module resolution broken on macOS with torch 2.9+ / Python 3.12+

Status: Open.

#21548 In sgl-project/sglang;

· karanb192 opened on Mar 27, 2026

[Bug] Incorrect output when using --enable-hisparse with NSA attention on DeepSeek-V3.2

Status: Open.

#21545 In sgl-project/sglang;

· Edward-lyz opened on Mar 27, 2026