yyj 61cb5cded5 fix(kt-kernel): restore MXFP4 decode fallback without AVX512 BF16 (#2196) ...

* fix(kt-kernel): restore MXFP4 decode fallback without AVX512 BF16

* Delete kt-kernel/test/per_commit/test_mxfp4_decode_bf16_guard.py

2026-09-10 23:14:05 +08:00

..

amx

fix(kt-kernel): restore MXFP4 decode fallback without AVX512 BF16 (#2196)

2026-09-10 23:14:05 +08:00

avx2

[feat](kt-kernel): packed-int4-resident AVX-VNNI-256 GPTQ INT4 backend (#2187)

2026-09-07 20:20:04 +08:00

kvcache

update kt-kernel

2025-11-03 15:19:52 +08:00

llamafile

[feat](kt-kernel): Ascend NPU backend for single-card DeepSeek-V4-Flash offload (#2157)

2026-09-01 20:10:24 +08:00

moe_kernel

[feat](kt-sft): Activation reuse & Int8 kernel refactor & native block-FP8 LoRA SFT (#2141)

2026-08-17 16:25:14 +08:00

sycl

[feat] Add SYCL backend for intel iGPU (#2089)

2026-07-17 16:24:04 +08:00

common.hpp

[feat](kt-sft): Activation reuse & Int8 kernel refactor & native block-FP8 LoRA SFT (#2141)

2026-08-17 16:25:14 +08:00

mla-tp.hpp

[feat](kt-sft): Activation reuse & Int8 kernel refactor & native block-FP8 LoRA SFT (#2141)

2026-08-17 16:25:14 +08:00

moe-sft-tp.hpp

[feat] Add native RAWINT4 routed-expert LoRA SFT for Kimi K2.5/K2.6 (#2186)

2026-09-10 17:05:16 +08:00

moe-tp.hpp

[feat](kt-sft): Activation reuse & Int8 kernel refactor & native block-FP8 LoRA SFT (#2141)

2026-08-17 16:25:14 +08:00

reduce.hpp

add kt-kernel

2025-10-12 05:13:00 +00:00

rms-norm.hpp

add kt-kernel

2025-10-12 05:13:00 +00:00

rope.hpp

add kt-kernel

2025-10-12 05:13:00 +00:00

sft_profile.hpp

[feat](kt-kernel): end-to-end full-parameter and LoRA SFT(#2094)

2026-07-23 21:23:34 +08:00

softmax.hpp

add kt-kernel

2025-10-12 05:13:00 +00:00

tp.hpp

add kt-kernel

2025-10-12 05:13:00 +00:00