Issues - kvcache-ai/ktransformers - Forgejo: Beyond coding. We Forge.

kvcache-ai/ktransformers

mirror of https://github.com/kvcache-ai/ktransformers.git synced 2026-11-05 01:07:28 +00:00

Search syntax

`assignee:<username>`	Filter by assigned user
`author:<username>`	Filter by the author
`is:all`	All
`is:closed`	Closed
`is:open`	Open
`mentions:<username>`	Filter by mentioned user
`modified:[>\|<]<date>`	Filter by last modified date
`review:<username>`	Filter by user who reviewed
`sort:<by>:[asc\|desc]`	Sort by: `created\|comments\|updated\|deadline`

New issue

[GH-ISSUE #2249] Quick survey: your experience as a PyTorch Ecosystem maintainer

#1312 opened 2026-10-29 14:51:07 +00:00 by bsakaguchi

[GH-ISSUE #2245] [Performance] Exact AVX2 FP8 arithmetic conversion avoids costly LUT gathers

#1311 opened 2026-10-29 14:51:07 +00:00 by bsakaguchi

[GH-ISSUE #2230] [Performance] NativeMoEWrapper rebuilds and retains a full checkpoint index per layer

#1310 opened 2026-10-29 14:51:06 +00:00 by bsakaguchi

[GH-ISSUE #2244] [Feature discussion] Full GLM-5.3 streamed FP8 GPU prefill on AVX2 / 2x32GiB SM120

#1309 opened 2026-10-29 14:51:06 +00:00 by bsakaguchi

[GH-ISSUE #2227] 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

#1308 opened 2026-10-29 14:51:06 +00:00 by bsakaguchi

[GH-ISSUE #2226] [Bug] TP2 FP8 layerwise prefill corrupts expert weights by reusing staging slots before all GPU readers finish

#1307 opened 2026-10-29 14:51:05 +00:00 by bsakaguchi

[GH-ISSUE #2218] RAWINT4 direct-pointer MoE frees safetensors mmap owners and segfaults on first inference (AVX2)

#1306 opened 2026-10-29 14:51:04 +00:00 by bsakaguchi

[GH-ISSUE #2213] Support for Deepseek V4.1 Flash

#1304 opened 2026-10-29 14:51:01 +00:00 by bsakaguchi 1 / 1

[GH-ISSUE #2208] A100 not support?

#1302 opened 2026-10-29 14:51:00 +00:00 by bsakaguchi 1 / 1

[GH-ISSUE #2184] cache_salt/extra_key are computed & carried but never delivered to Req — radix-cache tenant isolation is a no-op (regression vs upstream sglang)

#1299 opened 2026-10-29 14:50:59 +00:00 by bsakaguchi 1 / 1

[GH-ISSUE #2179] [Feature Request] qwen4_exp support (Qwen3.8-Flash-Next) — CPU/GPU hybrid expert offload via kt-kernel/sglang-kt

#1297 opened 2026-10-29 14:50:56 +00:00 by bsakaguchi 1 / 1

[GH-ISSUE #2172] [MXFP4 layerwise prefill] SM 8.6 (Ampere) excluded by capability whitelist — DSV4-Flash on dual A10 falls back to serialized full-GPU path

#1295 opened 2026-10-29 14:50:55 +00:00 by bsakaguchi

[GH-ISSUE #2151] GeneralMoEWrapper.load_weights() cpu_save path reads nonexistent dict keys, guaranteed KeyError

#1294 opened 2026-10-29 14:50:55 +00:00 by bsakaguchi 1 / 1

[GH-ISSUE #2150] kt-sft: chunked_prefill_size sized from cutoff_len, never scaled by batch — any per_device_train_batch_size>1 hard-errors

#1293 opened 2026-10-29 14:50:55 +00:00 by bsakaguchi

[GH-ISSUE #2127] DeepSeek-V4-Flash-0731 MTP failed to start on avx2 + sm89

#1285 opened 2026-10-29 14:50:51 +00:00 by bsakaguchi 1 / 1

[GH-ISSUE #2139] When will kvcache‑ai/sglang be synced with sgl‑project/sglang

#1284 opened 2026-10-29 14:50:51 +00:00 by bsakaguchi 1 / 1

[GH-ISSUE #2115] ROCm Support for Dual AMD Radeon Pro V620 (GLM-5.2)

#1283 opened 2026-10-29 14:50:50 +00:00 by bsakaguchi 1 / 1

[GH-ISSUE #2118] Support DSpark speculative decoding for DeepSeek-V4-Flash-0731 in sglang-kt

#1282 opened 2026-10-29 14:50:50 +00:00 by bsakaguchi 1 / 1

[GH-ISSUE #2113] MiniMax M2.7: AttributeError: 'LlamafileMoEWrapper' object has no attribute 'submit_write_weight_scale_to_buffer'

#1281 opened 2026-10-29 14:50:50 +00:00 by bsakaguchi 1 / 1

[GH-ISSUE #2109] kimi k3啥时候支持？

#1279 opened 2026-10-29 14:50:48 +00:00 by bsakaguchi 1 / 1

1 2 3 4 5 ...