Issues

• Quick survey: your experience as a PyTorch Ecosystem maintainer

  kvcache-ai/ktransformers#2249 · Filter by author michellearoth opened on Oct 16, 2026

• [Performance] Exact AVX2 FP8 arithmetic conversion avoids costly LUT gathers

  kvcache-ai/ktransformers#2245 · Filter by author f-wunderlich opened on Oct 12, 2026
  1

• [Feature discussion] Full GLM-5.3 streamed FP8 GPU prefill on AVX2 / 2x32GiB SM120

  kvcache-ai/ktransformers#2244 · Filter by author f-wunderlich opened on Oct 12, 2026

• [Performance] NativeMoEWrapper rebuilds and retains a full checkpoint index per layer

  kvcache-ai/ktransformers#2230 · Filter by author ZenAlexa opened on Oct 7, 2026

• Borrowed Prefill: lend the GPU's idle dense-weight VRAM to the prefill chunk — 3-4x faster hybrid prefill on 16 GB cards, same KV pool

  kvcache-ai/ktransformers#2227 · Filter by author 05yuki opened on Oct 7, 2026

• [Bug] TP2 FP8 layerwise prefill corrupts expert weights by reusing staging slots before all GPU readers finish

  kvcache-ai/ktransformers#2226 · Filter by author f-wunderlich opened on Oct 7, 2026
  2

• RAWINT4 direct-pointer MoE frees safetensors mmap owners and segfaults on first inference (AVX2)

  kvcache-ai/ktransformers#2218 · Filter by author f-wunderlich opened on Sep 30, 2026
  2

• Support for Deepseek V4.1 Flash

  enhancement
  kvcache-ai/ktransformers#2213 · Filter by author RobertoFMBarreto opened on Sep 21, 2026
  18

• A100 not support?

  enhancement
  kvcache-ai/ktransformers#2208 · Filter by author SunYong0821 opened on Sep 15, 2026
  1

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

  kvcache-ai/ktransformers#2184 · Filter by author wildoranges opened on Sep 2, 2026
  2

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

  enhancement
  kvcache-ai/ktransformers#2179 · Filter by author gdevenyi opened on Aug 28, 2026

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

  kvcache-ai/ktransformers#2172 · Filter by author 1stlook-dev opened on Aug 26, 2026
  2

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

  kvcache-ai/ktransformers#2151 · Filter by author AmirF194 opened on Aug 9, 2026
  1

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

  kvcache-ai/ktransformers#2150 · Filter by author PUMA-SPz opened on Aug 8, 2026
  1

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

  kvcache-ai/ktransformers#2139 · Filter by author stellarzhou opened on Aug 5, 2026

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

  kvcache-ai/ktransformers#2127 · Filter by author cadslab opened on Aug 3, 2026
  3

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

  enhancement
  kvcache-ai/ktransformers#2118 · Filter by author edwardyehuang opened on Aug 2, 2026
  1

• ROCm Support for Dual AMD Radeon Pro V620 (GLM-5.2)

  kvcache-ai/ktransformers#2115 · Filter by author DrSchnitzel57 opened on Jul 31, 2026

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

  kvcache-ai/ktransformers#2113 · Filter by author mrgaolei opened on Jul 31, 2026
  3

• kimi k3啥时候支持？

  enhancement
  kvcache-ai/ktransformers#2109 · Filter by author zhujianwei-ops opened on Jul 28, 2026
  1

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

  kvcache-ai/ktransformers#2108 · Filter by author AGenchev opened on Jul 27, 2026
  1

• RuntimeError: Only Tensors of floating point and complex dtype can require gradients

  kvcache-ai/ktransformers#2107 · Filter by author reinforceli opened on Jul 26, 2026

• Error with Deepskeep 4 flash with trtllm_fp4_block_scale_moe

  kvcache-ai/ktransformers#2106 · Filter by author ramon-garcia opened on Jul 23, 2026
  1

• [Bug] 无法正确识别模型文件

  kvcache-ai/ktransformers#2100 · Filter by author vibe-Chen opened on Jul 21, 2026

• kt-kernel/install.sh failed with A100

  kvcache-ai/ktransformers#2098 · Filter by author snailwei opened on Jul 20, 2026