Kingsley
349c56c51c
[data] Fix bugs of use_audio_in_video in Qwen2.5 Omni ( #7638 )
...
* cache _mm_inputs
* nit
* support for use_audio_in_video
* remove cache
* fix data
* Update mllm_video_audio_demo.json
2025-04-08 18:40:10 +08:00
Shawn Tao
acb09fa3a3
[trainer] fix key error ( #7635 )
2025-04-08 18:39:50 +08:00
Adarsh Shirawalmath
f75b91077b
[sglang] support transformers 4.51.0 ( #7639 )
2025-04-08 18:39:23 +08:00
hoshi-hiyouga
c3c0efbaa0
[misc] fix packing and eval plot ( #7623 )
2025-04-07 18:20:57 +08:00
hoshi-hiyouga
5115dc8c7f
[assets] update readme ( #7612 )
2025-04-06 13:58:49 +08:00
hoshi-hiyouga
831e7f1cfd
[model] add llama4 ( #7611 )
2025-04-06 13:42:31 +08:00
Kingsley
d4cfa9507e
[data] fix qwen2.5 omni plugin ( #7578 )
...
* specific entry
* Update mm_plugin.py
* fix fps cal
---------
Co-authored-by: hoshi-hiyouga <hiyouga@buaa.edu.cn>
2025-04-02 23:58:39 +08:00
Kingsley
d32c6c014d
[data] fix qwen2.5 omni plugin ( #7573 )
...
* align key with qwen2vl
* nit && change scripts
2025-04-02 21:28:52 +08:00
gechengze
7b9deb9410
[trainer] fix batch processing in PPO trainer ( #7576 )
2025-04-02 21:17:48 +08:00
hoshi-hiyouga
5e22597ff1
[infer] vllm video/audio inference ( #7566 )
2025-04-02 02:27:04 +08:00
hoshi-hiyouga
2bfcad2394
[model] fix kv cache ( #7564 )
2025-04-01 23:07:46 +08:00
Yu Shi Jie
a13b1bb49a
[model] fix use_cache patching for gemma3 multimodal ( #7500 )
2025-04-01 16:06:48 +08:00
Ritesh Goru
d10467d178
[data] specify position_ids in PackedSupervisedDatasetProcessor for neat_packing ( #7318 )
...
* use position_ids for neat_packing with fa2
* revert fa2 changes
2025-04-01 16:03:13 +08:00
taoharry
aac70663fd
[webui] fix launch with proxy ( #7332 )
2025-04-01 15:52:56 +08:00
Billy Cao
00409ff28a
[data] shard the dataset to allow multiprocessing when streaming is enabled ( #7530 )
...
* Shard the dataset when streaming to allow multiprocessing
* Allow user to not set dataset_shards to ensure backward compatibility
2025-04-01 15:36:23 +08:00
Hao
d70b3b4bc5
[trainer] new kto mismatch pair creation strategy ( #7509 )
2025-04-01 15:21:53 +08:00
hoshi-hiyouga
e76eba051d
[data] fix qwen2.5 omni collator ( #7553 )
2025-04-01 00:15:12 +08:00
Kingsley
7eed496336
[model] add Qwen2.5-Omni model ( #7537 )
...
* preserve image_sizes
* preserve image_sizes
* init plugin
* support audio-text2text lora
* nit
* support image/video-text2text, audio-text2text
* remove args
* remove lines
* add docs && nit
* remove some comments
* fix && add merge part script
* add license
2025-03-31 20:39:35 +08:00
hoshi-hiyouga
0f8296626a
[deps] pin pydantic to 2.10.6 ( #7546 )
2025-03-31 14:42:28 +08:00
Kingsley
8da1d2fa71
[data] fix pixtral plugin ( #7505 )
...
* preserve `image_sizes`
* add comments
2025-03-27 17:06:40 +08:00
Xu-pixel
b578a7d5b6
[3rdparty] support swanlab lark notification ( #7481 )
2025-03-27 01:52:01 +08:00
Kdump
24afceddb7
[trainer] fix wsd scheduler ( #7304 )
...
* [trainer] Warmup_stable_decay supports setting the number of stable and decay steps according to the warmup_ratio ratio
* Update trainer_utils.py
---------
Co-authored-by: hoshi-hiyouga <hiyouga@buaa.edu.cn>
2025-03-26 15:25:02 +08:00
hoshi-hiyouga
0583d06676
[model] add qwen2vl 32b & upgrade peft ( #7469 )
...
* add qwen2vl 32b
* fix ci
* upgrade peft to 0.15
* fix ci
* fix ci
2025-03-25 12:15:58 +08:00
GuoCoder
ec6a261568
[model] fix lora on quant models ( #7456 )
...
Co-authored-by: root <root@ai>
2025-03-25 11:59:46 +08:00
Xiaosu Zhu
6b3b97c738
[misc] update liger-kernel's monkey patch ( #7453 )
...
* Update liger_kernel.py
* Update setup.py
2025-03-25 11:58:52 +08:00
AbdelKarim ELJANDOUBI
6d3748f727
[misc] enable liger kernel for gemma3 text and paligemma ( #7466 )
...
* add gemma3 text
* add paligemma (1,2 and 2 mix)
2025-03-25 09:27:43 +08:00
Kenny Lam
7c890170e3
[misc] enable liger kernel for gemma3 ( #7462 )
2025-03-24 19:09:59 +08:00
hoshi-hiyouga
ca42c0c406
[assets] fix gemma3 readme ( #7449 )
2025-03-24 10:31:25 +08:00
hoshi-hiyouga
7203365b80
[trainer] fix vlm loss for transformers 4.49 ( #7448 )
2025-03-24 10:24:05 +08:00
rumichi
3612946dd9
[docker] upgrade to torch 2.6 ( #7442 )
2025-03-23 21:18:08 +08:00
hoshi-hiyouga
3aa4f32e9c
[misc] fix ci ( #7441 )
...
* fix ci
* improve ci
2025-03-23 21:09:35 +08:00
hoshi-hiyouga
304796b803
[misc] fix license ( #7440 )
2025-03-23 19:31:56 +08:00
SnowFox4004
7cfd6e4bb0
[scripts] support compute score on vllm's predictions ( #7419 )
...
* enable manual bleu&rouge eval by adding `scripts/eval_bleu_rouge.py`
* added libraries check
* update: 使用datasets库的多进程加速处理
* update:
- 使用 fire.Fire
- 修改代码格式
* Update eval_bleu_rouge.py: correctly uses fire
Deleted the code of using sys.argv
* Update eval_bleu_rouge.py
---------
Co-authored-by: SnowFox4004 <manba@out>
Co-authored-by: hoshi-hiyouga <hiyouga@buaa.edu.cn>
2025-03-23 19:21:01 +08:00
hoshi-hiyouga
05b19d6952
[deps] upgrade transformers to 4.50.0 ( #7437 )
...
* upgrade transformers
* fix hf cache
* fix dpo trainer
2025-03-23 17:44:27 +08:00
hoshi-hiyouga
919415dba9
[deps] upgrade vllm to 0.8 ( #7436 )
2025-03-23 14:32:22 +08:00
Guo, Quan
a959c2a509
[misc] fix sglang deps ( #7432 )
...
* feat: Add transformer version requirement for sglang
* feat: add srt to sglang which is required for running sglang
Other options are srt_hip, srt_xpu, srt_npu, srt_hpu, srt_cpu, for different computation architectures.
2025-03-23 14:07:10 +08:00
Eric Tang
db0a08db6f
[3rdparty] fix redundant process group destroy for ray ( #7395 )
...
* fix redundant process group destroy for ray
* Update tuner.py
---------
Co-authored-by: hoshi-hiyouga <hiyouga@buaa.edu.cn>
2025-03-21 10:56:47 +08:00
hoshi-hiyouga
a306f0f5a2
[version] fix minicpmo ( #7378 )
2025-03-20 16:59:31 +08:00
hoshi-hiyouga
63752fccf7
[assets] update wechat ( #7361 )
2025-03-18 21:31:09 +08:00
hoshi-hiyouga
1f9773395b
[misc] set dev version ( #7351 )
2025-03-18 00:10:53 +08:00
hoshi-hiyouga
128b5b12b3
[data] fix template ( #7349 )
2025-03-17 23:45:20 +08:00
hoshi-hiyouga
d5915a7dd7
[assets] update videos ( #7340 )
...
* Update README.md
* Update README_zh.md
2025-03-17 15:48:02 +08:00
Hertz
ec1154662b
[model] support hunyuan 7b ( #7317 )
...
* [Model]supported tencent-hunyuan model
* [Model]supported tencent-hunyuan model(fix)
* [Model]supported tencent-hunyuan model(fix)
2025-03-15 20:55:24 +08:00
Qiaolin Yu
a44a53ebec
[inference] support sglang backend ( #7278 )
...
* Mimic SGLang offline Engine
* Add more tests and args
* Pass all current tests
* Clean Code
* fix sample_params
* clean code
* Fix Stream Chat
* change sglang from engine mode to server mode
* fix
* Fix Review Issues
* Use SGLang Built-In Utilities
* Fix test SGLang
* Some Doc Issue
* fix sglang engine
* add readme
---------
Co-authored-by: Jin Pan <jpan236@wisc.edu>
Co-authored-by: hiyouga <hiyouga@buaa.edu.cn>
2025-03-15 04:37:58 +08:00
hoshi-hiyouga
93e6184cbe
[data] gemma3 plugin pan and scan ( #7294 )
...
* gemma3 pan and scan
* add test case
* fix test
2025-03-13 23:29:23 +08:00
hoshi-hiyouga
0be0d7796a
[assets] update video ( #7287 )
2025-03-13 18:45:47 +08:00
Ritesh Goru
480369a9f2
[data] efficient 4d_attention_mask creation in neat_packing ( #7272 )
2025-03-13 03:31:12 +08:00
hoshi-hiyouga
650a9a9057
[misc] update format ( #7277 )
2025-03-13 02:53:08 +08:00
hoshi-hiyouga
4b9d8da5a4
[model] support gemma3 ( #7273 )
2025-03-13 01:35:23 +08:00
hoshi-hiyouga
e6159ad730
[misc] upgrade deps ( #7257 )
2025-03-12 00:33:47 +08:00
