922 Commits

Author SHA1 Message Date
hoshi-hiyouga
3ce0fa0f39 Update issue templates
Former-commit-id: f967abcfcd052b65745f20e2c760ca45c412b66a
2023-11-16 18:56:30 +08:00
hiyouga
5a1409679f fix web ui demo
Former-commit-id: e566a68a27872f730b111078977048755ec74a40
2023-11-16 18:41:55 +08:00
hiyouga
22322a3fa1 fix web ui demo
Former-commit-id: 6fead193fe44fec74c2262d8653ed2f6006fac36
2023-11-16 17:12:23 +08:00
hiyouga
d05577134f release v0.3.0
Former-commit-id: de7f5b622340ab09ebbe57ad2703e63d06dfdeea
2023-11-16 16:00:11 +08:00
hiyouga
33302970f3 update readme
Former-commit-id: 4018aabc5d1623033d27a8aced25804de79b7e7b
2023-11-16 15:58:37 +08:00
hoshi-hiyouga
2990c68077 Merge #1525 from hiyouga/dev, fix #224 #336 #931 #936 #1011
Refactor llmtuner, support full-parameter RLHF

Former-commit-id: 3b92826803dc69471827b4f8204c2c3dc5310619
2023-11-16 15:47:13 +08:00
hiyouga
d9dc76530c fix css
Former-commit-id: 7afec127f60257462828298b25a5f6fd9c6f42c5
2023-11-16 15:45:38 +08:00
hiyouga
05478a0a2d fix bug in web ui
Former-commit-id: a598f145ec903dd2b2c984d951b6c450b142ece5
2023-11-16 15:21:24 +08:00
hiyouga
8e2e822c57 update ppo and demo in webui
Former-commit-id: de7571704c82121db13e3fc907379d2453100191
2023-11-16 14:55:26 +08:00
hiyouga
2b9ec24a5e fix bug in freeze tuning
Former-commit-id: f6b436a08421ca17d64abc51497f4aa43729a43b
2023-11-16 14:25:11 +08:00
hiyouga
2bd75dff0c tiny fix
Former-commit-id: d65519d8a44b73bbb713741c23465f13c35c83f5
2023-11-16 03:27:19 +08:00
hiyouga
e83b0cbafb fix rlhf callback
Former-commit-id: f5485452d660caef56474cb7dc37abbe4f34599e
2023-11-16 03:26:19 +08:00
hiyouga
71fe9ccdd4 fix bug in PPO training
Former-commit-id: 2e99f0e53ce6de0acbcab85dd50aef874e8c6336
2023-11-16 02:32:54 +08:00
hiyouga
77b1ed4deb fix import bug
Former-commit-id: 2356029cdd120d5f7bf630b80681ce8c53bff90d
2023-11-16 02:27:03 +08:00
hiyouga
685d0c975a support full-parameter PPO
Former-commit-id: 4af967d69475e1c9fdf1a7983cd6b83bd431abff
2023-11-16 02:08:04 +08:00
hiyouga
c970270c94 add demo mode for web UI
Former-commit-id: 5ad34f08b4e1505d7933b973497347f126b2e818
2023-11-15 23:51:26 +08:00
hoshi-hiyouga
96b87090c3 Create CODE_OF_CONDUCT.md
Former-commit-id: 6bee64cdf9c75488033e600fb5b48738daa1ed3b
2023-11-15 20:42:15 +08:00
hiyouga
dc16a3dd10 update readme and constants
Former-commit-id: 7d83e3dd9101a4fdd0b589d0c1f7b609c0feecd1
2023-11-15 18:04:37 +08:00
hiyouga
ddd0e61e57 support multiple modules in freeze training #1514
Former-commit-id: 60abac70dfd778df2ae8b3a2e960ed8b607d7ab6
2023-11-15 17:08:18 +08:00
hiyouga
d65b835835 fix imports
Former-commit-id: 6156f1abef631c675d150dd1cb0325cfc3820c91
2023-11-15 16:47:45 +08:00
hiyouga
5a206d54c9 disentangle model from tuner and rename modules
Former-commit-id: 02cbf91e7e424f8379c1fed01b82a5f7a83b6947
2023-11-15 16:29:09 +08:00
hiyouga
e2ad2be2ae fix #1507
Former-commit-id: 1ba9c53bd9743fa95fca1516c0ed9da352dbe9a1
2023-11-15 16:22:32 +08:00
hiyouga
d3b36079f1 Update cal_lr.py
Former-commit-id: b92ef6c80ae108982046ec1419efb67c8b10b250
2023-11-14 21:14:42 +08:00
hiyouga
3e4739253f Update cal_lr.py
Former-commit-id: b6c3f9b24324403db41c5680a00aabc6d53bbeb9
2023-11-14 21:13:01 +08:00
hiyouga
6ce2ef1b3d Update cal_lr.py
Former-commit-id: 1258eec806f6f4580a6eb7d9eb44f431f4c0da4f
2023-11-14 21:09:30 +08:00
hiyouga
3743205c0e add cal_lr.py
Former-commit-id: cea2ba17efc47917e63437a376f220864f7f90dd
2023-11-14 20:58:37 +08:00
hiyouga
72e69631b1 fix #1494
Former-commit-id: 07c8d734529f03e47ef638a1bda222e8824d3d38
2023-11-14 18:07:20 +08:00
hiyouga
3cb0fb4e11 fix #1489
Former-commit-id: ebdeaca9cdfd6138c690a0fcb9f676deaddff177
2023-11-14 15:27:05 +08:00
hiyouga
a834f566d8 support eval remote dataset
Former-commit-id: 71dd2698bf8c0b9ef7af995fb1e49e39fa66074e
2023-11-14 02:42:30 +08:00
hiyouga
4da322c225 fix dc link
Former-commit-id: 04c3a1f1c98d8f191102e359def0c8dcdc9621e3
2023-11-13 23:22:56 +08:00
hiyouga
17ae8f7a52 release v0.2.2, fix #1478 #1466
Former-commit-id: c9534c411716e1dceb54c5eb35fe845c93ee2973
2023-11-13 23:09:05 +08:00
hiyouga
088a31ff70 fix #424
Former-commit-id: ca24d445f825e120e659f5cd080a954c2243b8f2
2023-11-13 22:42:23 +08:00
hiyouga
c6dfbfa62c refactor evaluation, upgrade trl to 074
Former-commit-id: ed09ebe2c1926ffdb0520b3866f7fd03a9aed046
2023-11-13 22:20:35 +08:00
hiyouga
d88c72f326 fix flashattn warning
Former-commit-id: 6eb095d39bd82fdbdb729a0ea57fc7246e3a60d6
2023-11-10 18:34:54 +08:00
hiyouga
dc69e3025e add todo
Former-commit-id: 0bd884feb11736d0ab24ca19885151cb47d9dcd3
2023-11-10 14:38:18 +08:00
hiyouga
6a80ba8de4 refactor constants
Former-commit-id: a4d4c3fd35276f20e3b354e9d13ea971029c8775
2023-11-10 14:16:10 +08:00
hiyouga
3e3837f363 tiny fix
Former-commit-id: 97ba2027bb1ddc01a3c824c40d5a180828810c2c
2023-11-09 17:20:49 +08:00
hoshi-hiyouga
d5255674e0 Merge pull request #1454 from yyq/main
Update finetuning_args.py

Former-commit-id: e67d8b93705383a8590f99e26e9fe8f663712aef
2023-11-09 17:12:18 +08:00
Yanqing
3df8b4fe96 Update finetuning_args.py
更新 chatglm/falcon/bloom 的 lora_target 的名称

Former-commit-id: 06606739af035a80ae9ddba9d12c965ed289305d
2023-11-09 17:04:40 +08:00
hiyouga
f3c0dbfb26 fix #1452
Former-commit-id: 4d16214467715df458e24d03bb7d303d62b8bdcd
2023-11-09 16:41:32 +08:00
hiyouga
25ce19e8b4 update readme
Former-commit-id: f7ead54042868550a3e8a6928ea3c0e2673f15b3
2023-11-09 16:00:24 +08:00
hiyouga
cb263c0f45 release v0.2.1
Former-commit-id: 1c30f2be0140f5ab47c2bc811170d0271a0cdad6
2023-11-09 15:54:16 +08:00
hiyouga
f697474e67 add template, modify datasets
Former-commit-id: 81e54beb4d0f792f4fd7f450643caaf10f2f0b7d
2023-11-09 15:53:23 +08:00
hoshi-hiyouga
f16b254b28 Merge pull request #1436 from lvzii/main
fix tokenizer config changed after pretrain

Former-commit-id: f485c3983e413fd3a3a57b451800705b072869a7
2023-11-09 14:30:50 +08:00
hiyouga
5528d4be98 support parquet format #1446
Former-commit-id: 44a3b9ac9f10d2012b8ad3d8c48123db9a0da2f1
2023-11-09 14:17:40 +08:00
hiyouga
52f33dcdec fix #1438 #1439
Former-commit-id: 84260d58dda22adc32c26bc943ed2a36fd01341d
2023-11-09 13:45:10 +08:00
lvzi
90143468dd fix tokenizer config changed after pretrain
Changing tokenizer's attribute at preprocessing stage will result in saving a wrong tokenizer.
for example, baichuan2

Former-commit-id: 19942b5314b84267691f0a5657d0679f2ddbe58b
2023-11-08 15:50:46 +08:00
hiyouga
a3973795a0 fix ppo train and dpo eval
Former-commit-id: ced863031836632cb5920e22ae6991f251372118
2023-11-07 22:48:51 +08:00
hiyouga
cd619fef98 fix #1422
Former-commit-id: 25d7bbd0a5142f001bd2ff498df07b24137050a9
2023-11-07 19:42:01 +08:00
hiyouga
8e1c942844 fix reward model loading
Former-commit-id: 9709ca501180a1afce32e9043aedb359762b437d
2023-11-07 17:20:51 +08:00