423A35C7

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-05-21 15:38:56 +08:00

7e20db5735 [v1] support liger_kernel (#10493)

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-05-20 23:18:56 +08:00

2322bf1cc2 [v1] add cuda fused moe kernel, implementing with triton (#10481)

368c48968f [callback] add torch profiler callback (#10463)

8b5ea65770 [v1] support reward training stage (#10431)

Compare 3 commits »

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-05-20 06:58:55 +08:00

40e786d016 [data] add missing return statement in MiniCPM V Plugin (#10500)

423A35C7 synced commits to main at 423A35C7/pytorch3d from mirror 2026-05-15 12:38:54 +08:00

c307c64c70 Suppress type errors for Pyre upgrade

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-05-14 03:58:55 +08:00

6b9df75ab9 [docker] update npu docker (#10479)

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-05-12 19:18:55 +08:00

ca50f22c38 [fix] Fix MiniCPM-V-4.6 image preprocessing behavior (#10478)

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-05-09 01:28:55 +08:00

53e77a9bfa [model] support MiniCPM-V-4.6 (#10472)

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-05-08 17:18:55 +08:00

55bd4944b6 [fix] fix qwen3_6 template doc (#10470)

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-05-08 00:58:54 +08:00

7e09152275 fix(data/converter): handle None tool_calls in OpenAI-style messages (#10455)

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-05-07 08:38:56 +08:00

1e503a982d [assets] correct typo in examples/README_zh.md (#10462)

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-05-04 01:28:54 +08:00

8752280dd7 [data] Optimize QwenVL video dataset preprocessing (#10404)

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-05-01 16:18:55 +08:00

468723c5d9 [packing] fix GDN crash when meeting dummy image (#10453)

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-05-01 08:08:55 +08:00

887ee2b121 [refactor] Add KTransformers AMX MoE SFT support via Accelerate (#10430)

6b08b948c9 [misc] bump transformers version upperbound (#10446)

Compare 2 commits »

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-04-30 07:38:58 +08:00

f7f3bfcbd7 [model] support Hy3-Preview (#10432)

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-04-29 23:18:56 +08:00

3475198d1e [fa2] fix IMA when train qwen3_5 (#10448)

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-04-28 19:08:58 +08:00

50945ef850 [v1] fix device_mesh and sp for fsdp2 (#10429)

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-04-28 02:39:03 +08:00

2f0bef207a [export] handle NotImplementedError in export_model for transformers>=5.0 (fixes #10410) (#10438)

2092abc217 [npu] add Qwen3.5 support with Partial RoPE and Hybrid Attention (#10421)

Compare 2 commits »

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-04-27 18:29:08 +08:00

99464b3d03 [misc] code lint (#10439)

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-04-27 01:59:00 +08:00

9a0cfdccfa [v1] fix init on meta in transformers v5 (#10414)

c8890c32db [data] support discard history cot for multiturn (#10435)

79c8332e4c [train] add qwen35 patch for neat_packing (#10436)

Compare 3 commits »

423A35C7 synced commits to main at 423A35C7/LLaMA-Factory from mirror 2026-04-24 08:38:56 +08:00

e0bc3c1971 [v1] fix epoch and steps (#10422)