Commit Graph

  • 6d6e0f44fc
    [trainer] new kto mismatch pair creation strategy (#7509) Hao 2025-04-01 15:21:53 +08:00
  • 2d421c57bf
    [data] fix qwen2.5 omni collator (#7553) hoshi-hiyouga 2025-04-01 00:15:12 +08:00
  • 185c76f6ad
    [model] add Qwen2.5-Omni model (#7537) Kingsley 2025-03-31 20:39:35 +08:00
  • 468eea6f6d
    [deps] pin pydantic to 2.10.6 (#7546) hoshi-hiyouga 2025-03-31 14:42:28 +08:00
  • 49436e93e6
    [assets] update wechat (#7523) hoshi-hiyouga 2025-03-28 17:44:36 +08:00
  • b00cb2ed42
    [data] fix pixtral plugin (#7505) Kingsley 2025-03-27 17:06:40 +08:00
  • f547334604
    [3rdparty] support swanlab lark notification (#7481) Xu-pixel 2025-03-27 01:52:01 +08:00
  • 01166841cf
    [trainer] fix wsd scheduler (#7304) Kdump 2025-03-26 15:25:02 +08:00
  • 59e12bffe8
    [model] add qwen2vl 32b & upgrade peft (#7469) hoshi-hiyouga 2025-03-25 12:15:58 +08:00
  • b6d8749bf3
    [model] fix lora on quant models (#7456) GuoCoder 2025-03-25 11:59:46 +08:00
  • bc9ada9db7
    [misc] update liger-kernel's monkey patch (#7453) Xiaosu Zhu 2025-03-25 11:58:52 +08:00
  • b6dc7e01e2
    [misc] enable liger kernel for gemma3 text and paligemma (#7466) AbdelKarim ELJANDOUBI 2025-03-25 02:27:43 +01:00
  • 59a56f7226
    [misc] enable liger kernel for gemma3 (#7462) Kenny Lam 2025-03-24 11:09:59 +00:00
  • 9abee9cd1a
    [assets] update wechat (#7455) hoshi-hiyouga 2025-03-24 14:53:10 +08:00
  • 833edc7c73
    [assets] fix gemma3 readme (#7449) hoshi-hiyouga 2025-03-24 10:31:25 +08:00
  • 42e090d38b
    [trainer] fix vlm loss for transformers 4.49 (#7448) hoshi-hiyouga 2025-03-24 10:24:05 +08:00
  • 747e02d60d
    [docker] upgrade to torch 2.6 (#7442) rumichi 2025-03-23 22:18:08 +09:00
  • c841e92116
    [misc] fix ci (#7441) hoshi-hiyouga 2025-03-23 21:09:35 +08:00
  • fbf49e2500
    [misc] fix license (#7440) hoshi-hiyouga 2025-03-23 19:31:56 +08:00
  • 7d4dc25c23
    [scripts] support compute score on vllm's predictions (#7419) SnowFox4004 2025-03-23 19:21:01 +08:00
  • b1b78daf06
    [deps] upgrade transformers to 4.50.0 (#7437) hoshi-hiyouga 2025-03-23 17:44:27 +08:00
  • dfbe1391e9
    [deps] upgrade vllm to 0.8 (#7436) hoshi-hiyouga 2025-03-23 14:32:22 +08:00
  • ebc989ad4a
    [misc] fix sglang deps (#7432) Guo, Quan 2025-03-23 14:07:10 +08:00
  • d8a5571be7
    [3rdparty] fix redundant process group destroy for ray (#7395) Eric Tang 2025-03-20 19:56:47 -07:00
  • 555b71a1cb
    [version] fix minicpmo (#7378) hoshi-hiyouga 2025-03-20 16:59:31 +08:00
  • 4a5d0f0ba7
    [assets] update wechat (#7361) hoshi-hiyouga 2025-03-18 21:31:09 +08:00
  • c518146e62
    [misc] set dev version (#7351) hoshi-hiyouga 2025-03-18 00:10:53 +08:00
  • 1d2131e5cb
    [data] fix template (#7349) hoshi-hiyouga 2025-03-17 23:45:20 +08:00
  • 48a6584fb1
    [assets] update videos (#7340) hoshi-hiyouga 2025-03-17 15:48:02 +08:00
  • a71e685021
    [model] support hunyuan 7b (#7317) Hertz 2025-03-15 20:55:24 +08:00
  • 30038d9ce7
    [inference] support sglang backend (#7278) Qiaolin Yu 2025-03-14 16:37:58 -04:00
  • ef5f1c1def
    [data] gemma3 plugin pan and scan (#7294) hoshi-hiyouga 2025-03-13 23:29:23 +08:00
  • 3dff4ecca8
    [dataset] fix ultrachat_200k dataset (#7259) Victor Nogueira 2025-03-13 13:20:18 +01:00
  • 0dbce72fb8
    [assets] update wechat (#7288) hoshi-hiyouga 2025-03-13 18:48:59 +08:00
  • e9b427d535
    [assets] update video (#7287) hoshi-hiyouga 2025-03-13 18:45:47 +08:00
  • d7d79f7e06
    [data] efficient 4d_attention_mask creation in neat_packing (#7272) Ritesh Goru 2025-03-13 01:01:12 +05:30
  • 9ccfb97a2c
    [misc] update format (#7277) hoshi-hiyouga 2025-03-13 02:53:08 +08:00
  • 165d3ed084
    [model] support gemma3 (#7273) hoshi-hiyouga 2025-03-13 01:35:23 +08:00
  • 142fd7e755
    [misc] upgrade deps (#7257) hoshi-hiyouga 2025-03-12 00:33:47 +08:00
  • 7c1640ed5f
    [misc] upgrade format to py39 (#7256) hoshi-hiyouga 2025-03-12 00:08:41 +08:00
  • cdafa8a15e
    [ci] update workflow (#7255) hoshi-hiyouga 2025-03-11 22:57:49 +08:00
  • b256ca86f0
    [core] release v0.9.2 (#7254) hoshi-hiyouga 2025-03-11 22:42:23 +08:00
  • 7a7071e504 Merge pull request #7242 from hiyouga/hiyouga/release v0.9.2 hoshi-hiyouga 2025-03-11 15:28:45 +08:00
  • 847ae972d0 Merge pull request #7247 from hiyouga/hiyouga/commit hoshi-hiyouga 2025-03-11 15:28:04 +08:00
  • 1c634d9c53 Merge pull request #7244 from hiyouga/hiyouga/token hoshi-hiyouga 2025-03-11 15:17:15 +08:00
  • 99b71768a0 support commit info hiyouga 2025-03-11 15:13:59 +08:00
  • 37b844d929 remove exit in preprocess hiyouga 2025-03-11 15:06:17 +08:00
  • f5810a6e47 release v0.9.2 hiyouga 2025-03-11 14:48:22 +08:00
  • 317d0855d2 [infer] fix vllm args (#7235) hoshi-hiyouga 2025-03-11 01:15:35 +08:00
  • 0a43bc1960 [tracking] add swanlab_logdir param (#7219) Ze-Yi LIN 2025-03-11 00:53:07 +08:00
  • 5a29f49fb1 [config] update args (#7231) hoshi-hiyouga 2025-03-10 23:04:43 +08:00
  • 4e68828e46 [config] fix export max len (#7230) hoshi-hiyouga 2025-03-10 16:46:08 +08:00
  • 9a0044ef5e [assets] update wechat (#7229) hoshi-hiyouga 2025-03-10 15:39:06 +08:00
  • d412301d08 [data] update mm demo data (#7211) hoshi-hiyouga 2025-03-07 20:07:15 +08:00
  • 5a0fd22c05 [assets] update readme (#7209) hoshi-hiyouga 2025-03-07 17:27:49 +08:00
  • df63f05b47 [data] fix loader (#7207) hoshi-hiyouga 2025-03-07 17:20:46 +08:00
  • 98ea0e8109 [misc] fix ds config (#7205) hoshi-hiyouga 2025-03-07 15:21:28 +08:00
  • 33b4c33279 [data] fix function formatter (#7201) ZhangChuanhui 2025-03-07 15:17:23 +08:00
  • 113cc3d920 [misc] fix cli (#7204) hoshi-hiyouga 2025-03-07 15:01:18 +08:00
  • b6c0e8608e [script] fix vllm version (#7193) hoshi-hiyouga 2025-03-06 17:14:17 +08:00
  • eba31ae313 [webui] support escape html (#7190) hoshi-hiyouga 2025-03-06 16:52:21 +08:00
  • e7556b591e [deps] upgrade vllm (#7183) hoshi-hiyouga 2025-03-06 15:25:08 +08:00
  • 2b21c749c1 [data] fix mm template (#7181) hoshi-hiyouga 2025-03-06 15:18:32 +08:00
  • 002f58ef8e [model] add QwQ 32b (#7179) hoshi-hiyouga 2025-03-06 11:58:36 +08:00
  • c67d2b9327 [trainer] fix swanlab callback (#7176) Ze-Yi LIN 2025-03-06 00:33:37 +08:00
  • 6e58115f98 [trainer] update config (#7174) hoshi-hiyouga 2025-03-05 23:32:54 +08:00
  • 8dddffa340 [data] fix qwen2audio plugin (#7166) sirui.li 2025-03-05 18:03:36 +08:00
  • e1d574a784 [assets] update wechat (#7161) hoshi-hiyouga 2025-03-05 14:11:10 +08:00
  • caef0a8937 [data] use bicubic resampler (#7143) hoshi-hiyouga 2025-03-04 00:17:06 +08:00
  • 392533e139 [webui] fix webui (#7142) hoshi-hiyouga 2025-03-04 00:01:49 +08:00
  • 299cd03785 [data] bailing template (#7117) rabbit 2025-03-03 15:33:22 +08:00
  • ee1b580328 [inference] fix hf_engine (#7120) hoshi-hiyouga 2025-03-01 05:22:49 +08:00
  • 54a090079c [assets] update wechat (#7106) hoshi-hiyouga 2025-02-28 12:01:04 +08:00
  • 210cdb9557 [webui] display swanlab exp link (#7089) Ze-Yi LIN 2025-02-27 19:40:54 +08:00
  • e86cb8a4fa [npu] update cann base image and torch 2.4 (#7061) leo-pony 2025-02-25 23:32:01 +08:00
  • f4aa0a146c [misc] fix project toml (#7067) hoshi-hiyouga 2025-02-25 23:22:48 +08:00
  • 96636c3729 [script] add seed args (#7058) JieShen 2025-02-25 19:44:57 +08:00
  • 81947f1d2c [model] add paligemma2-mix series (#7060) Kingsley 2025-02-25 18:51:16 +08:00
  • dca5fe14c2 [data] fix mllama (#7053) hoshi-hiyouga 2025-02-24 22:05:38 +08:00
  • ca78ba964d [model] add models (#7054) hoshi-hiyouga 2025-02-24 22:05:13 +08:00
  • 9359ee18ad [assets] update readme (#7051) hoshi-hiyouga 2025-02-24 20:45:06 +08:00
  • 15f3087b96 [assets] update wechat (#7019) hoshi-hiyouga 2025-02-20 20:32:33 +08:00
  • 1fcedf9af6 [data] fix MiniCPMV plugin (#6998) Zhangchi Feng 2025-02-19 19:36:04 +08:00
  • b0bbacaacb [webui] update css (#6985) hoshi-hiyouga 2025-02-18 18:27:57 +08:00
  • beb1a9f9d9 [data] add r1 distill dataset (#6983) hoshi-hiyouga 2025-02-18 17:25:09 +08:00
  • 3fbd4848e8 [version] support transformers 449 (#6982) hoshi-hiyouga 2025-02-18 17:05:40 +08:00
  • 184c5d0882 [misc] fix script (#6977) hoshi-hiyouga 2025-02-18 17:00:46 +08:00
  • 1f4a0b11ba [data] update vlm args (#6976) hoshi-hiyouga 2025-02-18 02:12:51 +08:00
  • b1d31ff0f9 [data] add min resolution option (#6975) hoshi-hiyouga 2025-02-18 01:40:46 +08:00
  • a8c9d5663d [data] fix predict dataset (#6972) hoshi-hiyouga 2025-02-17 20:29:40 +08:00
  • 475a355b82 [assets] update wechat (#6963) hoshi-hiyouga 2025-02-17 15:23:17 +08:00
  • 3dc938268c [data] fix minicpmo template (#6946) Zhangchi Feng 2025-02-15 00:37:41 +08:00
  • e55ec42d3c [ray] specify ray storage path (#6920) Eric Tang 2025-02-14 05:55:41 -08:00
  • 2baf8bf03d [misc] fix lora regex (#6944) hoshi-hiyouga 2025-02-14 21:38:43 +08:00
  • 13e1b7ee2b [misc] fix grad ckpt (#6931) hoshi-hiyouga 2025-02-13 23:27:51 +08:00
  • cd493b91de [model] add liger kernel to qwen2_5 vl (#6930) hoshi-hiyouga 2025-02-13 23:05:54 +08:00
  • 48173b606c [trainer] fix gen_kwarg to eval during training (#5451) Billy Cao 2025-02-13 02:35:06 +08:00
  • 0ad9f7f058 [data] evaluate on each dataset (#5522) SrWYG 2025-02-13 02:19:03 +08:00
  • 1adb46875f [data] improve error handling (#6128) Noah 2025-02-13 01:39:41 +08:00
  • 9b852ebe25 [misc] update readme (#6918) hoshi-hiyouga 2025-02-13 01:01:41 +08:00