ktransformers · Issues· 510 open

Seems difficult to install

Intel Consumer Level AVX-VNNI Support

A100 not support?

GLM5.2 int8: 'AMXMoEWrapper' object has no attribute 'submit_write_weight_scale_to_buffer'

cache_salt/extra_key are computed & carried but never delivered to Req — radix-cache tenant isolation is a no-op (regression vs upstream sglang)

[Feature Request] qwen4_exp support (Qwen3.8-Flash-Next) — CPU/GPU hybrid expert offload via kt-kernel/sglang-kt

FAQs | 常见问题

[MXFP4 layerwise prefill] SM 8.6 (Ampere) excluded by capability whitelist — DSV4-Flash on dual A10 falls back to serialized full-GPU path

GeneralMoEWrapper.load_weights() cpu_save path reads nonexistent dict keys, guaranteed KeyError

DeepSeek-V4-Flash-0731 MTP failed to start on avx2 + sm89

kt-sft: chunked_prefill_size sized from cutoff_len, never scaled by batch — any per_device_train_batch_size>1 hard-errors

Error with Deepskeep 4 flash with trtllm_fp4_block_scale_moe

When will kvcache‑ai/sglang be synced with sgl‑project/sglang

MiniMax M2.7: AttributeError: 'LlamafileMoEWrapper' object has no attribute 'submit_write_weight_scale_to_buffer'

Support DSpark speculative decoding for DeepSeek-V4-Flash-0731 in sglang-kt