117 Commits

Author SHA1 Message Date
hiyouga
c5f28b1496 fix recursion error
Former-commit-id: c7938188c36a71a878bca982b7dd151195164986
2023-10-22 16:28:37 +08:00
hiyouga
31969e78f7 reimplement neftune
Former-commit-id: efe9e5a194d3a9f052701d904715238816e4c09e
2023-10-22 16:15:08 +08:00
anvie
202cd43027 add NEFTune optimization
Former-commit-id: 603e0298af64116ac07130fe6661a9ba823c186c
2023-10-21 13:24:10 +07:00
hiyouga
35882be86e fix #1218
Former-commit-id: b301f35bd4a3bf368159c8f5fb4e2736f922115b
2023-10-19 16:17:41 +08:00
hiyouga
d539e524fa fix #1217
Former-commit-id: 065fc0a6f3f005bb87e1c5c126c8b6bb470ce700
2023-10-19 15:52:24 +08:00
hiyouga
f872297409 fix config, #1191
Former-commit-id: 5dbc9b355e85b203cb43ff72589374f0e04be391
2023-10-15 18:28:45 +08:00
hiyouga
cb68572bce refactor export, fix #1190
Former-commit-id: 30e60e37023a7c4a2db033ffec0542efa3d5cdfb
2023-10-15 16:01:48 +08:00
hiyouga
8ee069d7c3 fix loading dtype
Former-commit-id: d54a356128f7e335c12089702cf3de7f5b4baf16
2023-10-14 20:15:24 +08:00
hiyouga
0c40610fb5 fix #1176 #1177
Former-commit-id: 5627a2b57c270a78095a32083e2dc7aa02162875
2023-10-14 20:00:17 +08:00
hiyouga
7fea444d48 fix #1184
Former-commit-id: 5b069a967823e659dbc70b0d50361b3ad248087e
2023-10-14 19:20:11 +08:00
hiyouga
19582ee70f fix ppo args
Former-commit-id: 0f12899951808f53a482082eb116bda309775930
2023-10-11 23:40:50 +08:00
hiyouga
f1a8fcf917 refactor model_dtype, fix PPO trainer
Former-commit-id: 3e17ee5afbcb823a7c9a2f91864b3750cd79edb4
2023-10-11 23:16:01 +08:00
hiyouga
0c1e00574d fix shift short attention
Former-commit-id: 9a49cce8e6f6b222f74a07bdab40efee6a77b0f1
2023-10-09 17:07:46 +08:00
hiyouga
b1bc191c45 update webui #1086
Former-commit-id: 65a48bc398f18f71f5f2659b2070e3b9593af243
2023-10-09 14:50:14 +08:00
hiyouga
de8a0d689c fix #1097
Former-commit-id: c5b8796322d9d48e815038f9fecf0ce39036a4ee
2023-10-08 22:29:26 +08:00
hiyouga
536c32d8d4 add llamafy_qwen.py
Former-commit-id: 6cdc91543c022edcc98076488f06e809fde9bad7
2023-10-08 22:05:36 +08:00
hiyouga
4617413bde fix layer norm dtype
Former-commit-id: 67af21961b68d9b54d07b09e444c7140869f26da
2023-09-28 00:25:55 +08:00
hiyouga
d1d5ecb403 fix #1026
Former-commit-id: d0940d0dbd03d4bbcc955304566b0d5507edf9e6
2023-09-27 22:57:09 +08:00
hiyouga
57dafc37cc fix #424
Former-commit-id: daaf89f1126112a73b9f115b0f5617a8cd974a3e
2023-09-27 22:49:43 +08:00
hiyouga
867e513e18 refactor finetuning Args
Former-commit-id: be425a70a4c8f051717cf1e4464dbd79dae4c0b5
2023-09-27 22:28:06 +08:00
hiyouga
d6f5a3cae9 support LongLoRA
Former-commit-id: 0832ed37e7947d699f17375648a52f80752c2b6b
2023-09-27 21:55:50 +08:00
hiyouga
e4542e1d45 add MMLU and C-Eval script
Former-commit-id: 3403f876127b4b99c5e3edb2834cc3b9a3a0063f
2023-09-23 00:34:17 +08:00
hiyouga
eb41bc4aae fix #1000
Former-commit-id: 85de2d0a99e4a81fae890a963ccbb5c6142d52d4
2023-09-22 15:00:48 +08:00
hiyouga
dd6e9b3cc1 tiny fix
Former-commit-id: d24ea58c1a44b94227f4cb60f13fc1dd79997d01
2023-09-21 19:52:06 +08:00
hiyouga
d8ab75ee44 fix #944
Former-commit-id: 032245647848aaa4167086636b6c985268c5fee3
2023-09-21 19:51:02 +08:00
hiyouga
62ac4527b4 tiny fix
Former-commit-id: 1a7ddd8c1d20dc251f53923bd0ab9f3f1031dd21
2023-09-21 15:25:29 +08:00
statelesshz
b07a921f3e support export model on Ascend NPU
Former-commit-id: 50f94e6d9d62c848db7a3db85fa999d67ddd9f04
2023-09-20 10:26:02 +08:00
hiyouga
037584adfc add tests.cal_flops.py
Former-commit-id: 47a119db6c6e937f6ed96f70e3cda6031b9fbd0d
2023-09-16 23:40:41 +08:00
hiyouga
2256514acb fix #913
Former-commit-id: d67c11d69277292648dd9889a7321345e2c0c437
2023-09-15 20:58:28 +08:00
hiyouga
b46a0af117 fix #887
Former-commit-id: e131bc03e05ccae3c6ad8bb42ccf2cdcc2cf3cea
2023-09-14 17:56:58 +08:00
mmbwf
0c3e98488a Update utils.py
Fix parameters load error.

Former-commit-id: 112850364c7fdb53e3a38d42861404fc519108ce
2023-09-14 15:38:04 +08:00
hiyouga
1b762cf22c fix ppo save model
Former-commit-id: 300ca6d904524f46cb520056e1319a1e9a13d169
2023-09-12 16:25:29 +08:00
hiyouga
df0b527c1e fix #762 #814
Former-commit-id: 9a30ee5009040afbc524dbac0dad99904b2adf5f
2023-09-12 16:10:10 +08:00
hiyouga
8afbd30695 update flashattn, fix ppo save model
Former-commit-id: 0b08bc3dac246d4aa3f89afb7172529dcad9c39f
2023-09-11 17:25:36 +08:00
hiyouga
f5689c0e6e remove PeftTrainer
Former-commit-id: cc0cff3e991f194732d278e627648e528118a719
2023-09-10 22:23:23 +08:00
hiyouga
f326178f89 support FlashAttention2
Former-commit-id: 23e56c5554b948d4f08ad87849b261eafd2c7890
2023-09-10 20:43:56 +08:00
hiyouga
7947b2f6cd fix #850
Former-commit-id: e5975c4c6b8bd47ec506b0d4a4703bee05495436
2023-09-10 14:22:03 +08:00
hiyouga
9a3b4d9ee0 fix lora target
Former-commit-id: d822e41e7ac7e310ee49e347fc45754284ce30b8
2023-09-09 17:04:45 +08:00
hiyouga
34da72ffbb support lora target auto find
Former-commit-id: bce9984733d88bf013847eed523d1c75fdf0995e
2023-09-09 15:38:37 +08:00
hiyouga
b4e2433e72 fix chatglm2 tokenizer
Former-commit-id: 1ab60b4a93fa1be5dfe6ffbd4deb64c0f9d9b431
2023-09-09 13:50:29 +08:00
hiyouga
9f52ff8cd6 fix bug in DPO data collator
Former-commit-id: 4fc262cdf1347691e253bdfbd96568db5a49c086
2023-09-08 20:45:07 +08:00
hiyouga
19dc1dbc64 fix #761
Former-commit-id: be76f6cbe5143f781b6b39603b80392253b3080a
2023-09-08 20:22:18 +08:00
hiyouga
801d1fa7b9 change to right-padding, update reward score #803
Former-commit-id: baa90415bc8f5ebd423d001378b51c3a3a6c2ec7
2023-09-08 20:04:31 +08:00
hiyouga
8b7eaee38d update requirements
Former-commit-id: d796a4a5709c390629bafbeb7c91fccf6a9076d0
2023-09-07 19:26:25 +08:00
hiyouga
970fd3944b fix #818
Former-commit-id: e81fd458c279ed2f3cee780e517482b425c8886d
2023-09-07 19:19:53 +08:00
hiyouga
52871b35b5 add deepspeed check in PPO training
Former-commit-id: e203ec7f71f504ccbaa89c27d20b8a0d9fa53f7e
2023-09-07 19:12:40 +08:00
hiyouga
c426dd6d69 fix #809
Former-commit-id: 2783ca75365d7c373cefba039788a48f0b8f35fc
2023-09-07 19:04:32 +08:00
hiyouga
47d9325873 fix baichuan templates
Former-commit-id: f48a49e835b32f3991cfad8874c7b9c78953809f
2023-09-07 18:54:14 +08:00
hiyouga
c5fcf5b3a5 refactor dataset_attr, add eos in pt, fix #757
Former-commit-id: 0feec9a830b917b36686b61938a66e842eccf930
2023-09-01 19:00:45 +08:00
codemayq
accf14aa9a add dataset stage check
Former-commit-id: 5c719a7ce988339d034a653456da9742dc2cec7c
2023-08-30 16:23:08 +08:00