hoshi-hiyouga
39169986ef
[trainer] fix pt loss ( #7748 )
...
* fix pt loss
* robust
* fix
* test
2025-04-17 03:15:35 +08:00
hoshi-hiyouga
86ebb219d6
[breaking] bump transformers to 4.45.0 & improve ci ( #7746 )
...
* update ci
* fix
* fix
* fix
* fix
* fix
2025-04-17 02:36:48 +08:00
Eric Tang
bb8d79bae2
[ray] allow for specifying ray.init kwargs (i.e. runtime_env) ( #7647 )
...
* ray init kwargs
* Update trainer_utils.py
* fix ray args
---------
Co-authored-by: hoshi-hiyouga <hiyouga@buaa.edu.cn>
2025-04-10 11:31:05 +08:00
hoshi-hiyouga
1abd71b551
[assets] update readme ( #7644 )
2025-04-09 01:06:06 +08:00
Shawn Tao
acb09fa3a3
[trainer] fix key error ( #7635 )
2025-04-08 18:39:50 +08:00
hoshi-hiyouga
c3c0efbaa0
[misc] fix packing and eval plot ( #7623 )
2025-04-07 18:20:57 +08:00
hoshi-hiyouga
831e7f1cfd
[model] add llama4 ( #7611 )
2025-04-06 13:42:31 +08:00
gechengze
7b9deb9410
[trainer] fix batch processing in PPO trainer ( #7576 )
2025-04-02 21:17:48 +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
hoshi-hiyouga
7203365b80
[trainer] fix vlm loss for transformers 4.49 ( #7448 )
2025-03-24 10:24:05 +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
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
650a9a9057
[misc] update format ( #7277 )
2025-03-13 02:53:08 +08:00
hoshi-hiyouga
264538cb26
[misc] upgrade format to py39 ( #7256 )
2025-03-12 00:08:41 +08:00
hiyouga
1890d3dafe
release v0.9.2
...
Former-commit-id: e7ed1782d4a006400de6fc0f864abd01f7fadeea
2025-03-11 14:49:13 +08:00
Ze-Yi LIN
18968405d0
[tracking] add swanlab_logdir param ( #7219 )
...
* feat: add swanlab_logdir param
* fix
Former-commit-id: 9215ad488b6ac6cd57fe8fa4acdacceb63f68ca5
2025-03-11 00:53:07 +08:00
hoshi-hiyouga
16419b2834
[data] fix loader ( #7207 )
...
* fix dataloader
* add test case
* fix type
* fix ci
* fix ci
* fix ci
* disable overwrite cache in ci
Former-commit-id: e84af0e140b1aafd1a6d6fe185a8e41c8fc5f831
2025-03-07 17:20:46 +08:00
hoshi-hiyouga
a255c3a476
[misc] fix cli ( #7204 )
...
Former-commit-id: 999f57133ca163c7108d2d5ee8194eca9b2109b4
2025-03-07 15:01:18 +08:00
hoshi-hiyouga
9f16c50155
[model] add QwQ 32b ( #7179 )
...
Former-commit-id: 8897e48b8cd55407812453ddd4ff98ac7bdc4e91
2025-03-06 11:58:36 +08:00
Ze-Yi LIN
25bb9f5ad9
[trainer] fix swanlab callback ( #7176 )
...
Former-commit-id: 6d9acf4bd30db24499118aee16bd19cb19ba9e3d
2025-03-06 00:33:37 +08:00
hoshi-hiyouga
7b985f55db
[trainer] update config ( #7174 )
...
Former-commit-id: 9f535d0e3c4ee3cd0f1b65218c2eee5d03f43c6f
2025-03-05 23:32:54 +08:00
Ze-Yi LIN
11672f760d
[webui] display swanlab exp link ( #7089 )
...
* webui add swanlab link
* change callback name
* update
---------
Co-authored-by: hiyouga <hiyouga@buaa.edu.cn>
Former-commit-id: 27a4b93871c63b839c92940766bd7e0177972c9b
2025-02-27 19:40:54 +08:00
Eric Tang
76f9bd1820
[ray] specify ray storage path ( #6920 )
...
Former-commit-id: 4be6b66b1eaa79955e936ce2b747a8837ecd1e49
2025-02-14 21:55:41 +08:00
Billy Cao
58e9ca8aa0
[trainer] fix gen_kwarg to eval during training ( #5451 )
...
* Correctly pass gen_kwarg to eval during model runs
* fix
* fix
---------
Co-authored-by: hiyouga <hiyouga@buaa.edu.cn>
Former-commit-id: 845d16122496311e08263610a6a922f82604de7b
2025-02-13 02:35:06 +08:00
marko1616
0c0cdc26bc
[trainer] fix llama3.2 vision kto train ( #6904 )
...
Former-commit-id: 1563e89adc8988fc6e4250634a3f1e385979b0e5
2025-02-12 19:09:14 +08:00
hoshi-hiyouga
86063e27ea
[data] fix ollama template ( #6902 )
...
* fix ollama template
* add meta info
* use half precision
Former-commit-id: 1304bbea69d8c8ca57140017515dee7ae2ee6536
2025-02-11 22:43:09 +08:00
hoshi-hiyouga
88eafd865b
[misc] support export ollama modelfile ( #6899 )
...
* support export ollama modelfile
* update config
* add system and num ctx
Former-commit-id: 8c2af7466f4015f300b51841db11bcd2505ebf20
2025-02-11 19:52:25 +08:00
hoshi-hiyouga
4d1791e905
[deps] upgrade vllm ( #6857 )
...
Former-commit-id: 4bd50f65a3d62528768561019fda2723d045c7fd
2025-02-08 15:02:28 +08:00
hoshi-hiyouga
c2022431aa
[misc] update license year & fix llama pro ( #6814 )
...
* fix llamapro script
* change year
Former-commit-id: d9ae594178796994d400a5f207d6499712816f89
2025-02-05 01:53:33 +08:00
hoshi-hiyouga
222423bcef
[breaking] support transformers 4.48 ( #6628 )
...
Former-commit-id: f154ab175c513a4d7bb866bf2cffc34b77b50508
2025-01-31 01:36:33 +08:00
yinpu
a8fae3869d
fix: avoid redundant normalization in DPO's SFT loss calculation ( #6722 )
...
Former-commit-id: 971a8ccbdacf130763d40c7ef82a711b2fc1292f
2025-01-21 13:38:02 +08:00
hoshi-hiyouga
7638f1070e
[optim] clean apollo ( #6645 )
...
* clean apollo code
* update readme
Former-commit-id: 38b8ec4a99189483124b54df9d6bc6b0d318855a
2025-01-15 01:42:50 +08:00
zhuHQ
c2120432db
[optim] add support to APOLLO ( #6617 )
...
Former-commit-id: 5a252e5a458457adbd19da3b68a3897ad2962824
2025-01-15 00:24:56 +08:00
hoshi-hiyouga
2a05941b14
[inference] fix stop token for object detection ( #6624 )
...
* fix stop token
* update minicpm data pipeline
* fix npu qlora examples
Former-commit-id: 844919fadaa8a61dfae47020971ea80730b2346f
2025-01-13 21:34:20 +08:00
hiyouga
647c51a772
imporve log
...
Former-commit-id: a6abf375975ffea3d51e1b944c9855b5f62ffac8
2025-01-08 09:56:10 +00:00
hiyouga
0ef1f981da
fix llamaboard with ray
...
Former-commit-id: bd8a432d6a980b1b24a551626304fe3d394b1baf
2025-01-07 09:59:24 +00:00
hiyouga
944a2aec4d
refactor ray integration, support save ckpt
...
Former-commit-id: 2f50b27e608b2092bfceab6c6e84e6631e973ee2
2025-01-07 09:39:10 +00:00
Eric Tang
4f31ad997c
run style check
...
Former-commit-id: 5ec33baf5f95df9fa2afe5523c825d3eda8a076b
2025-01-07 08:55:44 +00:00
Kourosh Hakhamaneshi
8683582300
drafting ray integration
...
Signed-off-by: Kourosh Hakhamaneshi <kourosh@anyscale.com>
Former-commit-id: 19c12ddae9350f6e25a270fe3372f5b9094cf960
2025-01-07 08:55:44 +00:00
hiyouga
d8bd46f1bf
fix #6546
...
Former-commit-id: 6fcf2f10faf3b1614896b091591eeef96d717e64
2025-01-07 06:30:44 +00:00
hiyouga
2aaf3697d7
fix #6499
...
Former-commit-id: dffc607220ff6dac15cf501ac9a3cdbe80c25211
2025-01-02 11:28:54 +00:00
hiyouga
f8f05a883b
fix #6482
...
Former-commit-id: 8577f52b4152efe6cc7a8b5f6d37b4f9ba6684e7
2024-12-30 06:03:07 +00:00
hiyouga
88b1874c04
fix #6448
...
Former-commit-id: 04f78e85af5af14b4c195936623e426a6a128af2
2024-12-27 16:54:39 +00:00
hiyouga
a897d46049
support report custom args
...
Former-commit-id: d41254c40a1c5cacf9377096adb27efa9bdb79ea
2024-12-21 21:42:45 +00:00
hoshi-hiyouga
0a869c4ed4
Merge pull request #6401 from Zeyi-Lin/hiyouga/swanlab
...
feat: add swanlab for experiment tracking and visualization.
Former-commit-id: e65fe507f7643bf40b0fc462805c7b7f8ef6b738
2024-12-21 14:09:33 +08:00
ZeYi Lin
8a41c96761
fix: by hiyouga suggestion
...
Former-commit-id: 41195f1bc69e4b5da7a265369d368b06754362cf
2024-12-20 16:43:03 +08:00
ZeYi Lin
e5d9d8c55d
feat: ui improve
...
Former-commit-id: 6a1effb1741a13ae5238b0e9b429b4cbe3b6534f
2024-12-20 11:03:02 +08:00
ZeYi Lin
925e421bde
fix: bugs
...
Former-commit-id: a2297f97f7587c77d55fbce9ffa81dc60d0b04a1
2024-12-19 21:08:16 +08:00
