81 Commits

Author SHA1 Message Date
hiyouga
384f0e7678 add bf16 lora option
Former-commit-id: 58e7d7ff0cf9bf30e53b3eb12576f38d31976413
2024-01-19 16:29:03 +08:00
hiyouga
a423274fd9 support function calling
Former-commit-id: 66533b3f65babf2429c92c0f8fafe4eff5e0ff63
2024-01-18 09:54:23 +08:00
hiyouga
4d6669c268 fix #1789
Former-commit-id: d86455f685fa531e651333e00b4fe54d895cf2e4
2024-01-09 18:31:27 +08:00
hiyouga
89f4ae51f9 fix #2127
Former-commit-id: 5a1aa33fa9b546ab520f0ba4cb9d996b87eb71ca
2024-01-09 14:49:13 +08:00
hiyouga
75fe1404b1 improve model export
Former-commit-id: 31255147a566a23ce1a48402662d14af8ac267ab
2024-01-05 18:51:49 +08:00
hiyouga
b460c9372f fix #2098
Former-commit-id: e62d9158cffbf1044396597ddaf15b1c0bc5f954
2024-01-05 17:11:26 +08:00
hiyouga
04ae80a52e fix #2081
Former-commit-id: ec4b539b6c0be11e15d273025c414b694bbd6c9a
2024-01-04 23:19:08 +08:00
hiyouga
8c74851b70 fix dispatch
Former-commit-id: deda82638716506dc690902c51276bb1eb0ddd5e
2024-01-03 16:33:16 +08:00
hiyouga
7168392a51 fix valuehead patch
Former-commit-id: d9cb98362b58b28ae0ee207e7c07e75e5d810876
2024-01-03 16:19:23 +08:00
hiyouga
ccc5b324fe fix rm server
Former-commit-id: 81bc1638682a9fd01518f9f25250a6b584d2a9e6
2024-01-03 15:30:46 +08:00
hiyouga
ebb32e85f8 fix version
Former-commit-id: dd7500b65d0d548441eece101b60d51fa619cc0f
2023-12-29 04:53:36 +08:00
hiyouga
c33fbea469 fix bug
Former-commit-id: b06faa1be3f5aa5e0fa31aa31314c213c36c3442
2023-12-24 19:20:12 +08:00
hiyouga
921f593632 update loader
Former-commit-id: 080d8eab858217ca58bffe719d5ffde7579c5bda
2023-12-24 19:10:23 +08:00
hiyouga
940403720a update patcher
Former-commit-id: d6d7b6670847ce4ea10353c5b126214542b45c2b
2023-12-23 15:24:27 +08:00
hiyouga
306a70c7ba fix unsloth dtype
Former-commit-id: fd22e6546ce5f38a6a075cf894aafc3d206b2fcd
2023-12-23 01:59:49 +08:00
hiyouga
d358d955e5 fix dpo trainer
Former-commit-id: c160dd7cd86e296e32775ace2e4258a473449c41
2023-12-23 01:51:55 +08:00
hiyouga
6faf9c35a9 support unsloth
Former-commit-id: b857f00234b90b785d82ca7cdb29af3d948b1a7b
2023-12-23 00:14:33 +08:00
ShaneTian
d05febe5de Fix slow model initialization in bfloat16 dtype.
Former-commit-id: cf2e2f6f9b7f09b1e2faf6fbc413e3f62e3846c7
2023-12-22 16:27:28 +08:00
hiyouga
79f301a2c6 fix ds zero3 check
Former-commit-id: 7f50705b1d821d287bd854211319f697f57b25db
2023-12-21 01:19:22 +08:00
hiyouga
31cbc67986 match version
Former-commit-id: 16db52522584a8e084d4db2a7c253c8b88f27371
2023-12-20 22:17:35 +08:00
hiyouga
3c330869ef improve quantization
Former-commit-id: 4dde60017ad8208dfea0b2bb61df6a14a35d03e0
2023-12-20 18:27:16 +08:00
hiyouga
dba1af4841 add max_memory for gptq #1923
Former-commit-id: 9afc42c8b999fbbc206d9a467ca5795b27a10096
2023-12-20 18:15:17 +08:00
hiyouga
de0bb1d2da fix mixtral inference #1821
Former-commit-id: 612f9fd19cbd29e8b1785a1576a9668e7dcd264c
2023-12-20 15:11:15 +08:00
hiyouga
cc16ece283 fix #1900
Former-commit-id: 4c35214396f873588562606b084740b6581188d9
2023-12-19 17:21:46 +08:00
hiyouga
790a31404a fix #1742
Former-commit-id: efbb32afdcf0d6aa4ca26f54c95f76dbb84f77dc
2023-12-16 20:50:45 +08:00
hiyouga
df777c30d1 add noisy mean initialization #1815
Former-commit-id: 3253b1fca0123071913079277186c160046edf21
2023-12-16 19:47:51 +08:00
hiyouga
d81ad2d4bc support dpo-ftx
Former-commit-id: 86dfa04f9821556019fa777106787f73eb70b452
2023-12-16 19:21:41 +08:00
hiyouga
9f77e8b025 support autogptq in llama board #246
Former-commit-id: fea01226703d1534b5cf511bcb6a49e73bc86ce1
2023-12-16 16:31:30 +08:00
hiyouga
296711d502 support quantization in export model
Former-commit-id: f32500ae6edccab7d14df4c92467e15986866def
2023-12-15 23:44:50 +08:00
hiyouga
9121722999 update dc link
Former-commit-id: f6789e50e17a377b6d9b434d8e12ad99d8eecfeb
2023-12-15 22:11:31 +08:00
hiyouga
33521fb45e fix bug
Former-commit-id: 95ac272907a04a64785f928536de1fd099150f92
2023-12-15 21:54:02 +08:00
hiyouga
e5204e60ed fix bug
Former-commit-id: 8b80baf02cfece53527c27712f0899fa3532c414
2023-12-15 21:49:26 +08:00
hiyouga
0409428d87 add configurer
Former-commit-id: c40c9889615ffb49c7ce24c69c0d3d20d841c800
2023-12-15 21:46:40 +08:00
hiyouga
f902b0d420 refactor adapter hparam
Former-commit-id: f82aece9ebd6df83a7a005cc7cbbcec07fa6e14d
2023-12-15 20:53:11 +08:00
hiyouga
27ef5b1aa7 add loftq
Former-commit-id: 0b900882ef19ac49604a24fbae8b3254f1bff7ad
2023-12-14 21:53:56 +08:00
hiyouga
c32303fc7e fix valuehead model
Former-commit-id: 9f628debb6510f2d1c91b00f121a721ab5d648e9
2023-12-14 20:15:20 +08:00
hoshi-hiyouga
45abe361ba tiny fix
Former-commit-id: 987df4c62f34026adfe2089910f4ff9ac6ebd9a6
2023-12-13 17:32:36 +08:00
hoshi-hiyouga
3ae479faae revert peft version
Former-commit-id: 6440fa1a8c28fd2db58d0905a67d071837e0edd1
2023-12-13 10:49:45 +08:00
hoshi-hiyouga
5698038f49 update peft version
Former-commit-id: 31c01e1272bd2cd9588e5ee68c1924a3dd55c67e
2023-12-13 10:23:51 +08:00
hoshi-hiyouga
020233f725 tiny fix
Former-commit-id: 1478bc052417e0939188f55a0adcbf00956960f2
2023-12-13 10:21:29 +08:00
hoshi-hiyouga
6f9d55b8eb fix #1819
Former-commit-id: f2e2b0354cbe9a7190ccab807f690cc8ab433a6e
2023-12-13 10:14:01 +08:00
hiyouga
2542b62d77 remove loftq
Former-commit-id: e175c0a1c631296117abda2403a4b87bbdd35a66
2023-12-13 01:53:46 +08:00
hiyouga
e39bbdd287 support loftq
Former-commit-id: e7ac2eb7f7daae17525a278ffbe2f82c0fbd8093
2023-12-12 22:47:06 +08:00
hiyouga
934d00ea1e support system column #1765
Former-commit-id: f425584a511c5e42bae8b3ba090eaa898b28adad
2023-12-12 19:45:59 +08:00
hiyouga
c27675f70d fix modelscope data hub
Former-commit-id: 5b63e8c22538a4788e4b6c8df50e6e6be93ceeac
2023-12-12 18:33:06 +08:00
hiyouga
9e2cc21d04 update readme
Former-commit-id: 42e042a4206aeb5177ddde56386e9655b0c06460
2023-12-12 11:44:30 +08:00
hiyouga
6975124a57 support mixtral
Former-commit-id: 75b5b8e36ab1933b2625f11b645f56cbc805fd85
2023-12-12 11:39:04 +08:00
hiyouga
9f69307db1 fix baichuan resize
Former-commit-id: 66956d13074a9bc74d7a737b9476f38361a7764a
2023-12-11 20:55:50 +08:00
hiyouga
c3448a045c tiny fix
Former-commit-id: 1f839fc4f278c2a258df22899241fc66a2cca682
2023-12-11 18:09:40 +08:00
hiyouga
95c561983c support resize embeddings #1786
Former-commit-id: 368a41bd3c6a04f869083058d9165954fbdad105
2023-12-11 17:50:02 +08:00