786 Commits

Author SHA1 Message Date
Johann-Peter Hartmann
7676ac5f5a Merge branch 'hiyouga:main' into main
Former-commit-id: 0395d0aafb69e86645e6b0a36b8f8dadb82219e0
2024-02-04 12:51:25 +00:00
hiyouga
274a61f39d fix reserved label len
Former-commit-id: b06d6c05a1911f329252a7572240048e456affdc
2024-02-04 17:54:26 +08:00
hiyouga
84ec6b5d01 fix #2420
Former-commit-id: 7a34087e4db62e603c9a9a26d8ff3910d7b10c40
2024-02-04 15:51:47 +08:00
hiyouga
f14cadf12d fix #2189
Former-commit-id: b3d81b229d376671e1c12229aeb487b0d84f2548
2024-02-04 00:47:37 +08:00
hiyouga
4adb4477bc bump up transformers version
Former-commit-id: 82f4d4301ed9f31b160d6313a1d2d44a22865f4d
2024-02-04 00:01:16 +08:00
hiyouga
dcca5901a6 fix #2397
Former-commit-id: 7404692808f2288d539668d364965ad104dacadb
2024-02-03 23:45:31 +08:00
hiyouga
80c0ad44a3 add hint for freeze #2412
Former-commit-id: 9600c93633629605573d908019563fa3870ad6f8
2024-02-03 23:38:56 +08:00
hiyouga
434e858c4e fix #2376
Former-commit-id: 8e2cfa7cca21b7fd4538d72114e36f704bcc82fe
2024-02-03 23:14:31 +08:00
hiyouga
fded83c00e support minicpm #2404
Former-commit-id: 4449e91cbee8fd804cf8bf1ff6b9f5301fde94ed
2024-02-03 22:36:46 +08:00
Johann-Peter Hartmann
d900ef1e23 add simple german chatml template chatml_de
Former-commit-id: 9f1d67c09f1af2c7aa383adec66842cacde92e33
2024-02-03 09:01:15 +01:00
Fallen Angel
e884171cd8 fix eos_token_id=0 bug
when eos_token_id=0, will never add eos_token

Former-commit-id: 576b4881c386d897462a875b28066ce9d6e06dd5
2024-02-02 17:34:48 +08:00
hiyouga
38228d2d9f fix #2388
Former-commit-id: 203a36c9adfd9aa0f35fbf8089c9138534d68c53
2024-01-31 17:23:56 +08:00
hiyouga
7f28b961a5 fix autoset attn impl, update data readme
Former-commit-id: 34a6e5f82baf45cc8dbb11f9f7ab4a480ab7ec5c
2024-01-31 11:58:07 +08:00
hiyouga
eafdae8e94 fix #2320
Former-commit-id: e0b0c4415aaf80e75f6dd4f3777a0616b0e60f84
2024-01-24 16:19:18 +08:00
ldwang
d1a35c3fb1 Add patch_mixtral_replace_moe_impl for full training Mitral using DeepSpeed Zero3.
Signed-off-by: ldwang <ftgreat@gmail.com>

Former-commit-id: 5f50c02f0e425737cd80abdf8fde9e25abf13083
2024-01-24 15:25:31 +08:00
ldwang
0df7de7ab0 Add patch_mixtral_replace_moe_impl for full training Mitral using DeepSpeed Zero3.
Signed-off-by: ldwang <ftgreat@gmail.com>

Former-commit-id: d1413dcec8a3b1d671f240b82a689c72b54d7b93
2024-01-24 14:43:16 +08:00
hiyouga
115e7af908 add hint
Former-commit-id: c540ef41bda61993b83ef8cfe3c84b1d169e984c
2024-01-22 23:32:01 +08:00
hoshi-hiyouga
7e27722cbb Merge pull request #2283 from A-Cepheus/main
fix: ZeRO3 does not work with MoE models
Former-commit-id: f5ea760abec2aac8d29ce5c945647be05648e676
2024-01-22 23:28:45 +08:00
hoshi-hiyouga
e3dfc7de11 Update patcher.py
Former-commit-id: 33556cc6b0b65cc6db02e66f4f6e75112c33d966
2024-01-22 23:27:39 +08:00
hiyouga
5db5ad21df fix #2282 and update tool prompt
Former-commit-id: 1c412f803866bde32b76f7c26c7b464b6b3651f3
2024-01-22 22:27:30 +08:00
hiyouga
956a654610 add orion models
Former-commit-id: a34db89d2a281d1a1ace29dfd5bd5d4ff7c2f657
2024-01-22 21:26:53 +08:00
A-Cepheus
fbaa6269d4 🐞 fix: typo
Former-commit-id: 57a3687ecd23237559aee0e8e811b782846f2415
2024-01-22 16:04:39 +08:00
A-Cepheus
ff4027e448 🐞 fix: typo, move MoE fix to patcher
Former-commit-id: 4ff28e99ff9b48df7150591c6bbd3723f22b7715
2024-01-22 16:01:58 +08:00
A-Cepheus
96f8e94cd4 fix: ZeRO3 does not work with MoE models
Former-commit-id: b2844c049a88ea89f8e1812e2d2e8662b4002965
2024-01-22 15:21:14 +08:00
hiyouga
1a6a77cb12 add array param format
Former-commit-id: bf910f8a5b21ee552fa9ab069610a3f5f611de57
2024-01-21 22:17:48 +08:00
hiyouga
eca9c306a2 fix api
Former-commit-id: cca004da28aaaa0788eaea62b83d3402b38a3011
2024-01-21 19:15:27 +08:00
hiyouga
e4fe8c79ee fix #2268
Former-commit-id: 300ecf9b9d7fd99fbb68f3d086e3ad973c2f894e
2024-01-21 14:11:38 +08:00
hiyouga
cb24733ce5 tiny fix
Former-commit-id: 66839ae94985ddfa13eca4542127119c919b9648
2024-01-21 13:26:12 +08:00
hoshi-hiyouga
43da8d1690 Merge branch 'main' into fix_export_model_dtype
Former-commit-id: 6c7d2729f28eb37a97820d73c05648eb7fb2ca87
2024-01-21 12:40:24 +08:00
hoshi-hiyouga
8cfcce504d Update tuner.py
Former-commit-id: 691420661f7115f809e76484c1f29f74637e7cd0
2024-01-21 12:39:38 +08:00
hoshi-hiyouga
867008e008 Merge pull request #2262 from fenglui/main
fix torch_dtype check of export_model

Former-commit-id: 37cacf73a534fed1b06b4f3c6724f3568ce095e3
2024-01-21 12:34:37 +08:00
hiyouga
b96c2e4e6a format
Former-commit-id: f28a1a0c1cdd0062ad7b6c2826f8ec107a200cff
2024-01-21 12:34:17 +08:00
yhyu13
6ac798eb2a Remove manully set use_cache; torch_dtype is not str, save model as bfloat16 used to fail;
Former-commit-id: 75557fb5df16fd6eda7586cf041a16822dcfee8e
2024-01-21 11:12:15 +08:00
Aleksandr
b85b5d70e6 add russian lang
Former-commit-id: f8ce6d75b56439027bb17ff4e59eeb9eb3b9bd34
2024-01-21 04:28:14 +03:00
fenglui
beed491242 fix torch_dtype check of export_model
Former-commit-id: 8813181b6bffa76e5c7cb1f4caceada611c54b9d
2024-01-21 05:01:53 +08:00
hiyouga
16ca00a7dc finish agent
Former-commit-id: d8d9d3afe32725fe79120fcd1a0970fdcdc45625
2024-01-21 01:47:33 +08:00
hiyouga
7cfacd896b fix api
Former-commit-id: a4149fbcd600d4f3815f9353e5e92c569719bed6
2024-01-21 00:03:09 +08:00
hiyouga
d249caa4aa fix internlm2 template
Former-commit-id: ae05b23eb86555dbfafc174aa6ceff736e7fc9fa
2024-01-20 23:33:50 +08:00
hiyouga
8a492c9038 fix cli_demo
Former-commit-id: e8336b3653f43618cf7cd70f8da004208de970c0
2024-01-20 23:27:10 +08:00
hiyouga
91bc080850 fix #2260
Former-commit-id: ba97550671811a27177306dd231bb427130b26fb
2024-01-20 23:22:09 +08:00
hiyouga
e40f06cd18 release v0.5.0
Former-commit-id: 602bb9b685009b9af234499be278404721542ac7
2024-01-20 20:21:39 +08:00
hiyouga
c0e4eebf17 format style
Former-commit-id: 53b683531b83cd1d19de97c6565f16c1eca6f5e1
2024-01-20 20:15:56 +08:00
hiyouga
01c22ad7f8 fix tests
Former-commit-id: 23f97bd437424ef43b2b84743d56acc5d1ca70d5
2024-01-20 19:58:04 +08:00
hiyouga
e5a751ded0 support longlora for main branch
Former-commit-id: f869501ad4c368df26534c41f62c6d63c6be17dd
2024-01-20 19:25:22 +08:00
hoshi-hiyouga
1ac46568ef Merge pull request #2201 from liu-zichen/token_embed_resize
support resize embed for zero3

Former-commit-id: c0d1b5e3aef70da6b115614bd1ed539a76d6547a
2024-01-20 17:45:38 +08:00
hiyouga
b3b8fc7492 add upcast_lmhead option
Former-commit-id: 7ef69a1697c11ff13e7503360e40ef36cfb1c345
2024-01-19 23:54:25 +08:00
hiyouga
c5a9f7f593 set use_reentrant=False
Former-commit-id: efa2e27d5ef6eaeb7baa7551c651ef10ab31400c
2024-01-19 23:29:54 +08:00
hiyouga
23fc4ac5cb fix #2249
Former-commit-id: 7ec64588c541422875adfdaf5692a27d05b96cb9
2024-01-19 21:44:32 +08:00
hiyouga
ec88dd2930 add bf16 lora option
Former-commit-id: 58e7d7ff0cf9bf30e53b3eb12576f38d31976413
2024-01-19 16:29:03 +08:00
hiyouga
f7b4261108 fix function formatter
Former-commit-id: 363a87376ad8fe4149b387f7ccd60f31f2a5fdf7
2024-01-18 16:01:07 +08:00