1470 Commits

Author SHA1 Message Date
hiyouga
13d7b48efe improve KTO impl., replace datasets
Former-commit-id: c450ee87a35ff9235f9b695b0de2e042b2971178
2024-05-18 03:44:56 +08:00
hoshi-hiyouga
97469892c3 Merge pull request #3785 from enji-zhou/feature/add_kto
add kto

Former-commit-id: 33a354548e78a7f7f51d63f80974920827d30252
2024-05-18 03:07:18 +08:00
hoshi-hiyouga
2d1583faba Merge pull request #3794 from jue-jue-zi/main
feat: pass the `max_lora_rank` parameter to vLLM backend
Former-commit-id: d7ff49f245cd34668cbe43366e5f1890876da5e7
2024-05-17 16:17:30 +08:00
hoshi-hiyouga
e4a2accf4a Update model_args.py
Former-commit-id: 964672745389e35580a7010b0f010bd5ee08d542
2024-05-17 16:16:41 +08:00
juejuezi
20326affde feat: pass the max_lora_rank parameter to vLLM backend
Former-commit-id: b20d62ba3ccc5c02529d19e22b7adcfe8b88c326
2024-05-17 16:07:39 +08:00
hiyouga
9af3dce3c8 add deepseek v2 lite model
Former-commit-id: 8af98176055b6fc28d16b03207b5abaa7de6104a
2024-05-17 13:25:36 +08:00
enji.zhou
03956053b8 add kto
Former-commit-id: db1d5a4f51faae61fe18666057353747b01f5b8d
2024-05-17 13:09:17 +08:00
hiyouga
1bbbcb5895 Update wechat.jpg
Former-commit-id: 84415492bfdc620507bff8c7a8eedbfba812ef51
2024-05-17 12:18:03 +08:00
hiyouga
947f0e9964 update badam example #3764
Former-commit-id: e5bba7cf1bd5317a2446b67ee5e0e245bb8b4ad4
2024-05-17 02:21:10 +08:00
hiyouga
780a1f5a4e better dtype handle in loading
Former-commit-id: d9f190ff1ea1cc4dd061e8b03d429caea037bca4
2024-05-17 02:14:56 +08:00
hiyouga
dfff5119b4 update examples
Former-commit-id: ddec9e1b842d407790637e9b0b181f8b26926db9
2024-05-17 01:02:00 +08:00
hiyouga
f4bf49e891 enable inbrowser in webui
Former-commit-id: 694a05fd044bbbad107ca8fed5494460c78e1981
2024-05-17 00:08:56 +08:00
hiyouga
22f71c152a add falcon 11b
Former-commit-id: d77bed4091a6a8fea682b39d3261e1e93dfe093f
2024-05-17 00:08:33 +08:00
hiyouga
5eb8107db2 fix examples #3769
Former-commit-id: 3df986c6793a51ec2cb5f31fd1808cd3a9883bc4
2024-05-16 19:12:09 +08:00
hiyouga
cae823ddf0 rename package
Former-commit-id: 308edbc4260d45907b4a9d3a45ec21d83e48aacb
2024-05-16 18:39:08 +08:00
hiyouga
93a289107b set dev version
Former-commit-id: b2fc7aeb03fbb40e9beb27e9958c958ee48e23cf
2024-05-16 02:17:31 +08:00
hiyouga
b5034f2b12 release v0.7.1
Former-commit-id: 1c910079d8544c433add2d949a8378822d1425c9
v0.7.1
2024-05-16 00:57:16 +08:00
hiyouga
6e6267f17c fix #3694
Former-commit-id: 2a67ab3925f0c17c4cb5e8c5a5e2cc6a9dc7d47e
2024-05-16 00:35:28 +08:00
hiyouga
a84f155563 fix #3606
https://github.com/huggingface/peft/pull/1706

Former-commit-id: 44cfa9a1cda4e7b2cefd7792d7c166971da2fd48
2024-05-15 23:05:02 +08:00
hiyouga
757e172509 add Yi-VL-34B model
Former-commit-id: a388cadfc0bf3f7197f265a925fe89598aa5ee0d
2024-05-15 22:58:19 +08:00
hiyouga
74727c03e8 add yi-vl 6b model
Former-commit-id: 73845fcc464a083d75e5dbe39d93611f1488ccfe
2024-05-15 20:02:41 +08:00
hiyouga
b4c5a08d06 fix yi vl vllm infer
Former-commit-id: 51d61fcc89a0acc6e17b97865e277845294c0bd3
2024-05-15 19:25:48 +08:00
hiyouga
7ebd06dc1a add NPU docker images
Former-commit-id: e1f4e53915fc4dcc309e2b1bea27f6d11f63083a
2024-05-15 19:20:11 +08:00
hoshi-hiyouga
82a10c569a Merge pull request #3748 from BUAADreamer/main
Add MLLM YI-VL and save processor config during training

Former-commit-id: 75f405ec30dff921e42c6c90b2722a0f8b26d41b
2024-05-15 16:40:54 +08:00
hoshi-hiyouga
e80e50805c Update visual.py
Former-commit-id: cbeef2aaea0577fd1929e7f156a2b8601b31814e
2024-05-15 16:39:57 +08:00
hiyouga
f2b4237db1 fix fsdp model loading
Former-commit-id: 008e3b3b1075199d1a62d510a8e0f212207a06b9
2024-05-15 16:32:28 +08:00
hoshi-hiyouga
e09d68985f Update patcher.py
Former-commit-id: 5a0c8a8d343adb15b510f65286ee08f33b1b2751
2024-05-15 15:37:07 +08:00
hoshi-hiyouga
3d65c4ceab Update template.py
Former-commit-id: 780ca8306b31d5ac856f68de3abed7e838848464
2024-05-15 14:20:39 +08:00
hoshi-hiyouga
cea8cea9dd Update trainer.py
Former-commit-id: aa4a8933dd520227401b7041dae40fc6fb2ddaa2
2024-05-15 14:13:26 +08:00
hoshi-hiyouga
7622300c4b Update workflow.py
Former-commit-id: c309605ff565dc34d043314269fce5881212c27c
2024-05-15 14:13:01 +08:00
BUAADreamer
e1c2ff41a0 rm extra import
Former-commit-id: db1622f76b0fe9d669af206299ecec10954647af
2024-05-15 12:48:18 +08:00
BUAADreamer
3f38ef9f59 cast dtype in mm_proj
Former-commit-id: d2bf69740043012a0025dd9d80c7adf979dc3a88
2024-05-15 11:22:15 +08:00
BUAADreamer
dbc7b1c046 modify style
Former-commit-id: 771bed5bde510f3893d12cafc4163409d6cb21f3
2024-05-15 10:18:10 +08:00
BUAADreamer
df3a974057 Merge branch 'main' of https://github.com/BUAADreamer/LLaMA-Factory
Former-commit-id: 3f4556454c3a9c8ae7db98081b88073fff790f15
2024-05-15 09:54:21 +08:00
BUAADreamer
7d1d73b941 Merge branch 'hiyouga:main' into main
Former-commit-id: 70461444991dc14536cbaa09905d619de8b3c7f4
2024-05-15 09:54:14 +08:00
BUAADreamer
92b184101f add yivl and save processor to model_dir
Former-commit-id: afc6c7b9fd350f9f611a220363a3caa930ac56aa
2024-05-15 09:54:00 +08:00
hiyouga
967b9c0a49 fix bug in vllm engine
Former-commit-id: 11bf282dcc0ee257f2c28f46cc1a8edcf62421dc
2024-05-15 02:17:54 +08:00
hiyouga
ef167f839d fix gen args
Former-commit-id: 144801db09ec7f183ab455d7a88c76de7639333d
2024-05-15 01:49:05 +08:00
hiyouga
213ba09b24 fix examples
Former-commit-id: 7e69e71a52c736d0e42afbf61a3b3c22db606bc2
2024-05-15 00:26:10 +08:00
hiyouga
c4743674ab update examples
Former-commit-id: 5bdad463875100e402329d47cd4c14bf9bc3b84b
2024-05-15 00:05:17 +08:00
hiyouga
be1114bb43 update readme
Former-commit-id: b96d84835f9237e7277bb86395e448348473d20f
2024-05-14 23:57:08 +08:00
hiyouga
943779eabc update readme
Former-commit-id: fc547ee591ef3cfc1bdbb8297a75a74f05c83c82
2024-05-14 23:55:49 +08:00
hiyouga
f5df1ceaf1 add npu examples
Former-commit-id: af343034dd31303be59678af9d1eae338864e884
2024-05-14 23:32:53 +08:00
hoshi-hiyouga
e32a44fe6b Merge pull request #3584 from zhou-wjjw/main
Enhancing Ascend 910A Training Efficiency in LlamaFactory with NPU

Former-commit-id: ee4752f6d209f3f8ac6cf90ef7304e26848e211b
2024-05-14 22:18:37 +08:00
hiyouga
ec9ed23cfd use robust envs
Former-commit-id: c187b20aaa0a0eb7300d537fd9006bf977a02854
2024-05-14 21:36:42 +08:00
hoshi-hiyouga
082506eba8 Update train.py
Former-commit-id: 1c3c4989022025db756965350ae0381fc9db32e5
2024-05-14 20:47:52 +08:00
hoshi-hiyouga
fe586de344 Apply suggestions from code review
Co-authored-by: Huazhong Ji <hzji210@gmail.com>
Former-commit-id: 9089bc70c8838cb80473e557a750855f7b7a7695
2024-05-14 20:44:21 +08:00
hoshi-hiyouga
332f44fa43 Apply suggestions from code review
Co-authored-by: Huazhong Ji <hzji210@gmail.com>
Former-commit-id: 0ac6e73f9971a9310026ddc609b5266cb1639b64
2024-05-14 20:44:04 +08:00
hiyouga
5a5d450648 fix #3728
Former-commit-id: cfaee8b4cf5f89d767a20a057d2335bd30ec83a2
2024-05-14 20:37:21 +08:00
BUAADreamer
6c1561d73c Merge branch 'hiyouga:main' into main
Former-commit-id: 60b99f80c2d40c0601fed1afdf6fe04c8401876f
2024-05-14 16:51:38 +08:00