104 Commits

Author SHA1 Message Date
hoshi-hiyouga
33525a34b6 fix qwen2 moe (#6684)
Former-commit-id: ab624419fa0ab23ef7a331a0ec14e393328772b5
2025-01-17 13:46:09 +08:00
hoshi-hiyouga
7638f1070e [optim] clean apollo (#6645)
* clean apollo code

* update readme

Former-commit-id: 38b8ec4a99189483124b54df9d6bc6b0d318855a
2025-01-15 01:42:50 +08:00
zhuHQ
c2120432db [optim] add support to APOLLO (#6617)
Former-commit-id: 5a252e5a458457adbd19da3b68a3897ad2962824
2025-01-15 00:24:56 +08:00
Zhangchi Feng
f7857c83e1 Support Inference of MiniCPM-V-2.6 and MiniCPM-o-2.6 (#6631)
* fix template name

* tiny fix

* support minicpm-o-2.6

* support inference of minicpmv

Former-commit-id: 7f3c64e853a7cdd49d02bf85e237611941ac7fa8
2025-01-14 17:34:58 +08:00
Zhangchi Feng
ae32c148d1 Support new features of MiniCPM-V (#6626)
* fix template name

* tiny fix

* support minicpm-o-2.6

Former-commit-id: 53034a61c7654358f46916cbc370910fb2aeff3b
2025-01-14 00:26:19 +08:00
codingma
11c38b9173 add nf4 qlora support on Ascend NPU (#6601)
* add nf4 qlora support on Ascend NPU

* add transformers version check

* add python>=3.10 requirement description for npu

* tiny fix

---------

Co-authored-by: hoshi-hiyouga <hiyouga@buaa.edu.cn>
Former-commit-id: 7912d1acac5f10dab22145fe729a90c57aad8d85
2025-01-13 19:43:36 +08:00
fzc8578
9dc7b6c7ac adapt to new mllm_param
Former-commit-id: 0775b71965863c2618c117726a1046a36d6d85b8
2025-01-11 00:16:34 +08:00
Zhangchi Feng
627548bf7f Merge branch 'main' into minicpmv
Former-commit-id: 8a9c90759feda975faadc5858bd44b7ea116e7fb
2025-01-11 00:01:36 +08:00
hiyouga
dc65ecdf09 refactor mllm param logic
Former-commit-id: b895c190945cf5d991cb4e4dea2ae73cc9c8d246
2025-01-10 15:45:48 +00:00
fzc8578
e63c2df0b1 fix some
Former-commit-id: cd5a1a8b9c6eb59d6e95f79573f60ad8668f1942
2025-01-10 20:27:06 +08:00
fzc8578
25d4889789 tiny fix
Former-commit-id: f088e580d3bacd0eecd0c3bf17e928eb49832ba1
2025-01-10 20:15:39 +08:00
Zhangchi Feng
8c0a721c4c Merge branch 'main' into minicpmv
Former-commit-id: d8840ae416660e23f1d615ffd404f519360151d9
2025-01-10 20:12:07 +08:00
fzc8578
9e972bc9ec add some
Former-commit-id: fede563aeb716ba5d1e368fd3e1182e4e580d248
2025-01-10 20:01:22 +08:00
hiyouga
647c51a772 imporve log
Former-commit-id: a6abf375975ffea3d51e1b944c9855b5f62ffac8
2025-01-08 09:56:10 +00:00
fzc8578
8c2a712247 add some
Former-commit-id: b4790c66c126567bd193de52a564e3ce11c94769
2025-01-06 19:32:39 +08:00
fzc8578
2c120aa0df add some
Former-commit-id: 81176fe226da89eace89cb202bad68e73b7c2a02
2025-01-04 11:11:15 +08:00
Yaser Afshar
8881237475 Add trust_remote_code parameter and remove True
- Introduced a new model parameter `trust_remote_code`
- Set the default value of `trust_remote_code` to `False`
  to enhance security


Former-commit-id: 4bf23f406cf5235c16f9f8139850c53354901814
2024-12-17 12:25:12 +00:00
hoshi-hiyouga
3d3324be5c Merge pull request #6364 from hiyouga/hiyouga/control_reenterent_gc
[model] support non-reenterent-gc

Former-commit-id: a8a13cb360980bb4acd493e33ed405e07460fe73
2024-12-17 19:58:36 +08:00
hiyouga
4196d5b4d6 support non-reenterent-gc & fix #6358
Former-commit-id: 20446141e408885eb36d512bfb2dfb62bbc0c20d
2024-12-17 11:41:59 +00:00
hiyouga
d670d62a66 generalized packing & fix #6343
Former-commit-id: 3b1e4194616cacd5c24f08b328e31a008bddcf29
2024-12-17 10:26:19 +00:00
hiyouga
bac2c64f87 support qwen2vl train proj only
Former-commit-id: 0e949ef03455726e907c6f1039e93ebe480c897a
2024-12-05 10:37:42 +00:00
hoshi-hiyouga
e3066d1489 fix visual patch
Former-commit-id: ac51fa37cc23518b30a6123e188964dce39be82f
2024-11-25 20:06:06 +08:00
hoshi-hiyouga
487e3f2507 fix #6136
Former-commit-id: b84e5d91a070c473ea820c379bf9b5abbca6df2c
2024-11-25 19:43:42 +08:00
hiyouga
bc4421eeef add forbidden modules
Former-commit-id: c9f4d051d0eca7515bab201afdef17f1ac1b3cb9
2024-11-23 18:34:15 +00:00
hiyouga
5003820a6a fix inputs
Former-commit-id: 7d535bb8cdf7e81edda81152e63c8cfe6c9dcc9f
2024-11-23 18:26:02 +00:00
marko1616
cd2485f28d Linter.
Former-commit-id: 719d124f65ebb18ba0a1212751da9909160fb6f1
2024-11-23 16:09:04 +00:00
marko1616
918a367378 Tiny fix.
Former-commit-id: 4c1cef12d812832eed58b5da562ba083104756d3
2024-11-23 16:09:01 +00:00
hiyouga
093eda2ad6 support rank0 logger
Former-commit-id: 84528eabe560091bfd866b6a0ca864085af7529b
2024-11-02 18:31:04 +08:00
hiyouga
ba66ac084f update tests
Former-commit-id: 4e92b656e324725048d914946e70867be20032ff
2024-11-02 12:41:44 +08:00
hiyouga
8185eb1890 fix incorrect loss value for vlms
Former-commit-id: 0aa29a71ce958343a2086090d647eb63b8f5f5be
2024-10-30 08:56:46 +00:00
hoshi-hiyouga
16a9a44849 Update visual.py
Former-commit-id: 6f1db7b9abfbdea1781452388d66df3e9f9a5dd9
2024-10-29 22:10:29 +08:00
Kingsley
42aa9c65be Merge branch 'hiyouga:main' into pixtral-patch
Former-commit-id: 438302edfdb66b6397266b8b17ac66f60a89300c
2024-10-29 21:01:25 +08:00
hiyouga
248d5daaff use pre-commit
Former-commit-id: 7cfede95df22a9ff236788f04159b6b16b8d04bb
2024-10-29 09:07:46 +00:00
hiyouga
8f5921692e update requires
Former-commit-id: cae0e688ddcead370821e126c192bddc53ff6017
2024-10-29 16:10:07 +08:00
Kingsley
be519c84d9 Merge branch 'hiyouga:main' into pixtral-patch
Former-commit-id: 2076d00dfbe1279a91207157fd6d9a118427626a
2024-10-08 21:04:08 +08:00
hiyouga
b2dc6dc59a tiny fix
Former-commit-id: d8ddd07c2ed14d871fb25743c20265fc99e3e221
2024-10-08 17:48:56 +08:00
Kingsley
1272cb00df Merge branch 'hiyouga:main' into pixtral-patch
Former-commit-id: 9372ac93f304db438383d539ccd00bffe7415dbc
2024-10-01 00:52:31 +08:00
hiyouga
20ee1d2e19 fix #5542
Former-commit-id: cf28e7418c2eb07e86923a53ef832ef218e45af1
2024-09-30 23:28:55 +08:00
Kingsley
cbc1dd0c88 sync with former
Former-commit-id: f8707e52586182144c4fb70c7c0de8bf7044ef5e
2024-09-30 20:27:05 +08:00
Kingsley
8fd84c375e fix some errors due to inconsistency of model cards
Former-commit-id: dd83265b9b8768eb8732f59ace128dfe4aac1c47
2024-09-30 19:58:34 +08:00
Zhangchi Feng
8b9a814653 Merge branch 'main' into pixtral-patch
Former-commit-id: 0cf52d48fbc505e2fba29e5df0f2e6722db7ac79
2024-09-30 12:37:03 +08:00
BUAADreamer
a769d0e3d4 fix constants
Former-commit-id: 69309a23598995aa1937fd8d80732a018c18db87
2024-09-29 22:00:01 +08:00
BUAADreamer
3cc5408da7 fix style
Former-commit-id: dc1bdcb69e6f2c605a2c533dab15613affc902f4
2024-09-29 21:39:37 +08:00
Zhangchi Feng
689f5c4554 Merge branch 'main' into main
Former-commit-id: 7566589b820e6030269523e9d08c312594f893ae
2024-09-29 21:32:54 +08:00
BUAADreamer
ab5d042cd3 add more llava-next series template
Former-commit-id: 93f64f2aebf41582d39aa8a2c6059e562ca694b0
2024-09-29 21:29:29 +08:00
BUAADreamer
85a919b6f7 fix readme
Former-commit-id: 867e7e70dbff207dbd78668af09a638654937f71
2024-09-29 20:45:02 +08:00
BUAADreamer
722e01c8ab fix some
Former-commit-id: aeca8c0f978cb9754e0526b40cd431aaf867044f
2024-09-29 17:55:40 +08:00
hoshi-hiyouga
1ded3abdf1 Update attention.py
Former-commit-id: 2adf79c195053bb4541e0317573a2c89da28b5bc
2024-09-29 10:47:41 +08:00
Kingsley
e641f1215a Tiny fix
Former-commit-id: ae66e1a545f4cd209a57fd824f9bfb7e94436cba
2024-09-29 00:00:23 +08:00
Amirreza A
ca736bcab7 made a small change to a warning about fa2 for gemma2 models.
Former-commit-id: e0695a026d822c896cb4f5b33e0c4f88441d75e9
2024-09-28 19:03:36 +03:30