Search results

GLM5.2 on RTX 6000 Pro - Loads but segfault on warmup inference

Status: Open.

#2050 In kvcache-ai/ktransformers;

· trilog-inc opened on Jun 17, 2026

unauthenticated pickle.loads on TCP ZMQ socket allows pre-auth RCE in balance_serve scheduler

Status: Open.

#2042 In kvcache-ai/ktransformers;

· AAtomical opened on Jun 16, 2026

部署Qwen3.6-35B-A3B-FP8出现的问题

Status: Open.

#2040 In kvcache-ai/ktransformers;

· yzr23 opened on Jun 12, 2026

# [Bug/Workaround] FlashInfer & Triton JIT compilation failure on heterogeneous GPUs (e.g., RTX 3090 + 4090) in TP mode

Status: Open.

#2039 In kvcache-ai/ktransformers;

· y8nqd4jtfn-cell opened on Jun 11, 2026

It seems that GLM 4.5/4.6/4.7 is not yet supported?

Status: Open.

#2038 In kvcache-ai/ktransformers;

· kcqyyxx opened on Jun 9, 2026

KT +SGLANG是否支持GGUF MOE模型读取，例如QWEN3.6 35B A3B GGUF 的 UNSLOTH量化

Status: Open.

#2034 In kvcache-ai/ktransformers;

· tech112233445566 opened on Jun 5, 2026

small model and colab t4

enhancementNew feature or requestNew feature or request

Status: Open.

#2024 In kvcache-ai/ktransformers;

· kmnnmk212-source opened on May 26, 2026

H200 部署 GLM 5.1

Status: Open.

#2019 In kvcache-ai/ktransformers;

· OldSixOne opened on May 21, 2026

昇腾NPU在最新版本里面的支持

enhancementNew feature or requestNew feature or request

Status: Open.

#2012 In kvcache-ai/ktransformers;

· superxc opened on May 18, 2026

按照适配NPU的文档无法完成kt的安装

Status: Open.

#2011 In kvcache-ai/ktransformers;

· SongCQi opened on May 18, 2026

Crash when enable MTP on KT 0.6.2 DeepSeekV4Flash

Status: Open.

#2009 In kvcache-ai/ktransformers;

· mrgaolei opened on May 16, 2026

Ambiguous startup warning wrt. sglang / kt-kernel / moe integration

Status: Open.

#2008 In kvcache-ai/ktransformers;

· usrlocalben opened on May 16, 2026