288 Commits

Author SHA1 Message Date
hiyouga
38505ae9e1 update accelerate ver for schedule_free optimizers
Former-commit-id: bdde35fd2e4a919c1d63ebfc9a0ea8ba0c97e14c
2024-09-09 22:51:08 +08:00
hiyouga
2bb1ee3292 fix mm plugin
Former-commit-id: 6dc2b00fa2bcbdbad71cb0d96e14d75a7db932e2
2024-09-09 22:41:28 +08:00
hiyouga
c52eeb70e7 fix qwen2vl preprocess
Former-commit-id: c93795ae14b1d5a0a3440d18f8197fd53cd013da
2024-09-09 22:33:33 +08:00
hiyouga
7ccb86b215 add docstrings, refactor logger
Former-commit-id: 54c69059379d77dc9046c144cbe2d0253de3a4da
2024-09-08 00:56:56 +08:00
hiyouga
995491594d tiny fix
Former-commit-id: 76f2e5950483c669a15a961f0554442b6eb5c4a6
2024-09-05 23:41:16 +08:00
hiyouga
3aa6a3e45b add e2e tests
Former-commit-id: 94d5b1bd8f49dabeb9e3c53d634cfb3c06b0241d
2024-09-05 21:52:28 +08:00
hoshi-hiyouga
de277a8ab8 Merge pull request #5372 from LDLINGLINGLING/main
增加了对minicpm3.0的适配'

Former-commit-id: 12743562639ccc6eb0caf170e7123d9844e2b4a6
2024-09-05 21:35:42 +08:00
liudan
1797fe50a4 根据代码规范修改了代码
Former-commit-id: 3d3fbaaff98da327e10bdebb4aedbdf1ec9565e8
2024-09-05 20:17:55 +08:00
hiyouga
4fccc65579 support Yi-Coder models
Former-commit-id: 359ef8bb0ebb8ccf9651ac2b737c5a705dab6bad
2024-09-05 03:12:24 +08:00
hiyouga
9df7a26e6b video datasets
Former-commit-id: 8cafc7b055a854f483ad1c67f3d487ffd34b5f89
2024-09-05 02:04:17 +08:00
liudan
09cff03026 增加了对minicpm3.0的适配'
Former-commit-id: d7ba97be484bf781d6fe80252ea29eb505b261bb
2024-09-04 23:10:05 +08:00
hiyouga
d5ea05cfff update get template
Former-commit-id: dabad5570bf4a6b1044c963d8f27717030f373ef
2024-09-04 22:36:20 +08:00
hoshi-hiyouga
1dfd1aaf82 Merge pull request #5323 from naem1023/feat/add-dataset-map-batch-size-argument
Add batch size of map function in the preprocessed dataset

Former-commit-id: 8f441c2b3a5bb84dec2c037a541084c0201726c6
2024-09-04 22:09:36 +08:00
hoshi-hiyouga
8ac74c8ccb fix #5228
Former-commit-id: 44d6947e554cd61cff23c297248fff32a5f554da
2024-09-04 19:10:30 +08:00
hiyouga
a3d47818b7 fix #5344
Former-commit-id: d41d43a7c37cd10e34c9f399d1a346ffaee641c3
2024-09-04 03:06:06 +08:00
hiyouga
22deca0e9e lazy image load
Former-commit-id: 47ea97fb1ba77de2e8a561904aa8fdc27c3f5025
2024-09-04 02:27:08 +08:00
hiyouga
f837ae8cb5 fix #5338
Former-commit-id: 69d0acacc38dc90a97ce18447f865f525fff086a
2024-09-03 17:45:17 +08:00
naem1023
46695e42cc feat: add batch size of map function in the preprocessed dataset
Former-commit-id: 209313eeeab8d1a7c320bd9aa90a5f4656082b7c
2024-09-02 13:52:47 +09:00
hiyouga
f13e974930 tiny fix
Former-commit-id: 3a6f19f017539a2be31e348845a8ed8712275dcb
2024-09-02 01:33:22 +08:00
hiyouga
859823eb23 add image num check
Former-commit-id: ce8c5a264742161101da2094d0fc31c7fa7fb641
2024-09-02 01:31:36 +08:00
hiyouga
bfdcc6bacf add rlhf-v dataset
Former-commit-id: 8e49940746c1a6ff910f07dbefbec14af9d0f3c6
2024-09-01 22:57:41 +08:00
hiyouga
413a206652 fix bug
Former-commit-id: 64cb947c60398dfdfc2877f898147b0240089ea3
2024-09-01 21:07:49 +08:00
hiyouga
cb776752f6 fix mixed mm inputs and rlhf-v
Former-commit-id: 9967ccb3aef3ca557ad6eafb78c6c99866857008
2024-09-01 20:52:47 +08:00
hiyouga
09a2ecebc4 add test mm plugin
Former-commit-id: a2a8c0b92c49fb1ee65de271aec651e011dcabc4
2024-08-31 01:53:38 +08:00
hiyouga
92c398166d tiny fix
Former-commit-id: bee1bd43b946501690d70e4980205f9d82404296
2024-08-30 03:21:50 +08:00
hiyouga
a83756b5e9 refactor mm training
Former-commit-id: 3382317e32f88ed377d3e7759bdeaf0f2559d22a
2024-08-30 02:14:31 +08:00
hoshi-hiyouga
5730243179 fix bug
Former-commit-id: a8f22d8895c07ce37e8cb34b7014f3e9bcdbf6a3
2024-08-30 02:05:26 +08:00
simonJJJ
d9fd8f26cc update
Former-commit-id: 734e019cc1c120c23305d6f6bc1fb5b8e0d21b30
2024-08-28 20:22:46 +08:00
simonJJJ
8a09b1e732 initial-commit
Former-commit-id: aeb85f200bd824748008dae6047c2607dfcdf174
2024-08-28 16:51:35 +08:00
hoshi-hiyouga
a7604a95c1 Merge pull request #5156 from YeQiuO/main
fix Llama-template's system prompt bug

Former-commit-id: 15be2963477e7d3a9fe4330d7701d457dc49b583
2024-08-20 00:09:03 +08:00
hoshi-hiyouga
103132aa99 Update template.py
Former-commit-id: ec72eeca521ba4ec71f0c52de9eec49da2cf0feb
2024-08-20 00:03:33 +08:00
hoshi-hiyouga
a921505f59 Update template.py
Former-commit-id: 5f3300ec5de564df23c94ebd9662c86708f37ddb
2024-08-19 23:40:16 +08:00
Huiyu Chen
66a7f4f128 Add SailorLLM template
Former-commit-id: 2502833a7755d653e8492cb7f1215dc0105b6ee0
2024-08-15 15:10:14 +08:00
“Wzw”
3e159a0a83 fix Llama-template's system prompt bug
Former-commit-id: bcbbf4506300fc132e68a39a9a6dfa5e61497c8b
2024-08-12 19:22:12 +08:00
hiyouga
b5146facff follow #5115
Former-commit-id: c87023d539875cd8e622d40212a5627c9c182fb8
2024-08-09 18:03:00 +08:00
hoshi-hiyouga
397e4daa5d Merge pull request #5115 from YeQiuO/main
fix: `Train on the last turn only` truncate bug
Former-commit-id: 51542cb15fea785d445ecf80bbad0364ebc0cb77
2024-08-09 17:58:27 +08:00
hoshi-hiyouga
54f57fb354 Update template.py
Former-commit-id: 4f62e1cb243d996d1764c3c86ca234847ad2c022
2024-08-09 16:27:42 +08:00
“Wzw”
13e5fff97a mask_history args verify valid
Former-commit-id: 2fa1e0b2add60142c178e5e21ebaad7132fa5b00
2024-08-08 10:12:01 +08:00
“Wzw”
0bd25c3a6b fix mask_history tiny bug
Former-commit-id: b5ca86cc07d38cf342e351aab16cce4319245792
2024-08-08 10:09:33 +08:00
moontidef
733cb9087b fix: fix the deepseekcoder template to avoid repeat problem
Former-commit-id: b82ecbedd0fecd85195217916cba3c21998bd10b
2024-08-05 23:55:45 +08:00
hoshi-hiyouga
0ac7824822 Merge pull request #4892 from piamo/main
update deepseek template

Former-commit-id: 8a2846cfe1266bedde84da175805dac78439c20e
2024-07-26 11:49:34 +08:00
hiyouga
d63beb7a24 fix #4928
Former-commit-id: 091010492b2afc8ad25359323aba760523a4c5af
2024-07-24 17:00:29 +08:00
hiyouga
8d06679a3f fix flashattn + packing
Former-commit-id: 4135e69406c09f3892276e4fee7e5757a52a42d6
2024-07-21 17:07:45 +08:00
huangpan.foo
ee4c3f32d1 update deepseek template
Former-commit-id: 44e48e2b82929888b0880c00519102da4eb38ca8
2024-07-19 15:02:54 +08:00
hiyouga
34f16cc635 follow #4878 fix #4684
Former-commit-id: 779aae83d253de0a86201ff87543b5d695e28d23
2024-07-18 22:06:12 +08:00
Shiyu Zhang
249adacc4d 仅仅训练最后一轮对话
Former-commit-id: 1e7b396ff2489055574fd3365425d26360d73897
2024-07-18 15:30:25 +08:00
hoshi-hiyouga
2b22a7da48 Merge pull request #4691 from codemayq/feature-suppot-eval-dataset
add eval dataset support

Former-commit-id: 15b399a82f45b08fc07d2957884fb7821eba9fd9
2024-07-15 01:00:34 +08:00
hoshi-hiyouga
30a3c6e886 Update preprocess.py
Former-commit-id: df52fb05b1b08887288bbaab7c612b7ac27c2290
2024-07-15 00:55:36 +08:00
hoshi-hiyouga
eed7cbb453 Update parser.py
Former-commit-id: 84e4047f8a1f78256be65f3f7bddce358ed9e882
2024-07-15 00:55:21 +08:00
hoshi-hiyouga
5633c0ab1e Update data_utils.py
Former-commit-id: 97a0e291c79f145950b54a11d03d81ada4784d22
2024-07-15 00:54:34 +08:00