kyy/llm_trainer
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
2,260 Commits 1 Branch 0 Tags
2df2be1c47aded0132b5cc86acd3926dca585bc1
Commit Graph
1527 Commits
Author SHA1 Message Date
hoshi-hiyougaandGitHub 5f3300ec5d Update template.py 2024-08-19 23:40:16 +08:00
Ricardo 384ab8db84 _is_bf16_available judgment supports npu 2024-08-16 02:58:22 +00:00
ZxillyandGitHub dc36fcc3de fix: report correct device count for intel xpu 2024-08-15 08:30:43 +00:00
Huiyu Chen 2502833a77 Add SailorLLM template 2024-08-15 15:10:14 +08:00
liu-zichen ddee718b31 fix lr not change 2024-08-13 16:33:34 +08:00
“Wzw” bcbbf45063 fix Llama-template's system prompt bug 2024-08-12 19:22:12 +08:00
hiyouga dc770efb14 add qwen2 math models 2024-08-09 20:20:35 +08:00
hiyouga e2a28f51c6 add adam_mini to readme 2024-08-09 20:02:03 +08:00
hoshi-hiyougaandGitHub ef482394f0 Merge pull request #5095 from relic-yuexi/feat-optimizer 
Feat optimizer
 2024-08-09 19:51:33 +08:00
hiyouga c87023d539 follow #5115 2024-08-09 18:03:00 +08:00
hoshi-hiyougaandGitHub 51542cb15f Merge pull request #5115 from YeQiuO/main 
fix: `Train on the last turn only` truncate bug
 2024-08-09 17:58:27 +08:00
hoshi-hiyougaandGitHub 4f62e1cb24 Update template.py 2024-08-09 16:27:42 +08:00
“Wzw” 2fa1e0b2ad mask_history args verify valid 2024-08-08 10:12:01 +08:00
“Wzw” b5ca86cc07 fix mask_history tiny bug 2024-08-08 10:09:33 +08:00
moontidef 82bc15dc79 feat: add support for adammini 2024-08-07 10:08:22 +08:00
moontidef 40908a36fa fix: rename optimzer to optimizer 2024-08-07 10:05:01 +08:00
moontidefandGitHub 55f32dfbf9 Merge branch 'hiyouga:main' into main 2024-08-06 00:18:45 +08:00
moontidef b82ecbedd0 fix: fix the deepseekcoder template to avoid repeat problem 2024-08-05 23:55:45 +08:00
hiyouga b7ca6c8dc1 fix #5048 2024-08-05 23:48:19 +08:00
codingma dc09d454f2 support gemma-2-2b 2024-08-01 13:45:48 +08:00
hoshi-hiyougaandGitHub cd420c1938 Merge pull request #5010 from Eruly/main 
Add Korean web UI (llamafactory-cli webui)
 2024-07-30 01:55:54 +08:00
eruly 371009e522 Add Korean web UI (llamafactory-cli webui) 2024-07-29 13:47:13 +00:00
codingma 2c1ca9f742 fix pissa save 2024-07-29 10:44:34 +08:00
hoshi-hiyougaandGitHub 8a2846cfe1 Merge pull request #4892 from piamo/main 
update deepseek template
 2024-07-26 11:49:34 +08:00
hoshi-hiyougaandGitHub 9839c6d1f6 Merge pull request #4950 from liuwwang/main and fi 
fix: Repair the issue where quantization failed after merging the adapter.
 2024-07-26 11:48:56 +08:00
hiyouga 30f8149d11 update webui 2024-07-24 21:11:51 +08:00
hiyouga ace1d44857 tiny fix 2024-07-24 18:33:39 +08:00
hiyouga 091010492b fix #4928 2024-07-24 17:00:29 +08:00
hiyouga 1550fe7331 add mistral nemo model 2024-07-24 16:25:53 +08:00
hiyouga 26533c0604 add llama3.1 2024-07-24 16:20:11 +08:00
LiuwwandGitHub f91a9a250a fix: Repair the issue where quantization failed after merging the adapter. 2024-07-24 14:31:29 +08:00
hiyouga 5665062ca0 tiny fix 2024-07-22 21:10:15 +08:00
hoshi-hiyougaandGitHub 26082fc6c9 fix #4917 2024-07-22 11:28:31 +08:00
hiyouga c333e2f49d tiny fix 2024-07-22 00:06:03 +08:00
hiyouga 4135e69406 fix flashattn + packing 2024-07-21 17:07:45 +08:00
huangpan.foo 44e48e2b82 update deepseek template 2024-07-19 15:02:54 +08:00
hiyouga 88c7fc1599 set dev version 2024-07-19 02:01:46 +08:00
hiyouga 8f6995081c update parser 2024-07-19 01:36:39 +08:00
hiyouga bbd5a64423 release v0.8.3 2024-07-19 01:21:18 +08:00
hiyouga 608de799a2 add unittest 2024-07-19 01:06:27 +08:00
hiyouga 779aae83d2 follow #4878 fix #4684 2024-07-18 22:06:12 +08:00
Shiyu Zhang 1e7b396ff2 仅仅训练最后一轮对话 2024-07-18 15:30:25 +08:00
hiyouga beec77a089 fix metrics #4786 2024-07-17 00:47:00 +08:00
hiyouga d774b94f12 support batch_eval_metrics, fix #4826 2024-07-17 00:33:00 +08:00
hiyouga bda302fbfb tiny fix 2024-07-15 23:09:50 +08:00
hoshi-hiyougaandGitHub 38bc411d42 Merge pull request #4821 from codemayq/feature-eval-split 
add "split" as suffix in eval task name
 2024-07-15 22:59:44 +08:00
hiyouga fd8cc49008 fix #4820 2024-07-15 22:32:07 +08:00
codingma 645211dc01 1. change the task name format 
2. delete split param in data_args.py
 2024-07-15 09:55:33 +08:00
hiyouga 99ab7a8c1c allow computing rouge in training 2024-07-15 01:16:26 +08:00
hiyouga 29ebcd75d5 fix up 2024-07-15 01:04:56 +08:00