All issues

Search results

[Enhance] Control token should not be treated as special token in user messages

Status: Open.

#4960 In InternLM/lmdeploy;

· HAOCHENYE opened on Sep 13, 2026

[Feature] 现在lmdeploy能支持模型GLM-5.3-Flash-W4A16-MTP运行吗？

Status: Open.

#4958 In InternLM/lmdeploy;

· simonjhy opened on Sep 12, 2026

[Bug] 0.16.0和0.17.0版本lmdeploy加载InternVL3-78B模型W4A16量化版时内存要比0.15.0以及之前的版本要多150GB内存

Status: Open.

#4957 In InternLM/lmdeploy;

· 18610361898 opened on Sep 12, 2026

[Bug] Slower decode speed in v0.17 when compared to v0.14

Status: Open.

#4953 In InternLM/lmdeploy;

· Viji2029 opened on Sep 10, 2026

[Bug] TurboMind SIGSEGV crash-loop triggered by session_len truncation path (deterministic libc offset across 3 crashes)

Status: Open.

#4944 In InternLM/lmdeploy;

· zambalee opened on Sep 8, 2026

[Bug] input_embeddings are omitted from prefix-cache identity, causing wrong KV reuse

Status: Open.

#4933 In InternLM/lmdeploy;

· wildoranges opened on Sep 3, 2026

[Bug] Mooncake external KV keys omit KV format and weights lineage, with no default tenant isolation

Status: Open.

#4930 In InternLM/lmdeploy;

· wildoranges opened on Sep 3, 2026

[Feature] 咱们最新版本的lmdeploy可以加载这个 Qwen3.8-Flash吗？

Status: Open.

#4905 In InternLM/lmdeploy;

· simonjhy opened on Aug 26, 2026

[Bug] PyTorch engine: Qwen3.5 hybrid-GDN + AWQ checkpoint crashes on TP>1 - 'AwqLinear' object has no attribute 'block_size'

Status: Open.

#4899 In InternLM/lmdeploy;

· u5084757454-collab opened on Aug 24, 2026

[Feature] 什么时候turbomind能够开始支持MTP或者DFlash之类的加速功能？

Status: Open.

#4889 In InternLM/lmdeploy;

· simonjhy opened on Aug 21, 2026

[Bug] pytorch backend silently hangs under concurrent decoding with qwen3_5_mtp speculative decoding

Status: Open.

#4883 In InternLM/lmdeploy;

· matrix72c opened on Aug 19, 2026

[Feature] 当前qwen-next等SSM类模型prefix cache缓存命中率低

Status: Open.

#4879 In InternLM/lmdeploy;

· Tsundoku958 opened on Aug 18, 2026