mPLUG-DocOwl icon indicating copy to clipboard operation
mPLUG-DocOwl copied to clipboard

DocOwl1.5 training code?

Open coder4nlp opened this issue 1 year ago • 12 comments

When will the training code be released?Thx.

coder4nlp avatar Apr 16 '24 05:04 coder4nlp

Hi, @coder4nlp , the training code is scheduled for release at the end of this month. If you are urgent to finetune our model, you can refer to the training code of mPLUG-Owl2 and make some revisions to adjust to our model. Some hyper-parameters can refer to our paper.

HAWLYQ avatar Apr 16 '24 05:04 HAWLYQ

Hi, @coder4nlp , the training code is scheduled for release at the end of this month. If you are urgent to finetune our model, you can refer to the training code of mPLUG-Owl2 and make some revisions to adjust to our model. Some hyper-parameters can refer to our paper.

almost there!

whalefa1I avatar Apr 29 '24 03:04 whalefa1I

Hi, @coder4nlp , the training code is scheduled for release at the end of this month. If you are urgent to finetune our model, you can refer to the training code of mPLUG-Owl2 and make some revisions to adjust to our model. Some hyper-parameters can refer to our paper.

almost there!

training codes with DeepSpeed is under debugging and testing 。゚・ (>﹏<) ・゚。

HAWLYQ avatar Apr 29 '24 03:04 HAWLYQ

Hi, @coder4nlp , the training code is scheduled for release at the end of this month. If you are urgent to finetune our model, you can refer to the training code of mPLUG-Owl2 and make some revisions to adjust to our model. Some hyper-parameters can refer to our paper.

almost there!

training codes with DeepSpeed is under debugging and testing 。゚・ (>﹏<) ・゚。

@HAWLYQ So sad......。゚・ (>﹏<) ・゚。

coder4nlp avatar May 01 '24 12:05 coder4nlp

@HAWLYQ can you test for deepspeed stage 3 integrations, specifically for deadlock issues while training/fine-tuning?

AR-javis avatar May 02 '24 06:05 AR-javis

@HAWLYQ can you test for deepspeed stage 3 integrations, specifically for deadlock issues while training/fine-tuning?

Hi, @AR-javis , I'm debugging with deepspeed zero3~ I will try my best to release it within a week~

HAWLYQ avatar May 02 '24 06:05 HAWLYQ

where are the schedules?

coder4nlp avatar May 06 '24 06:05 coder4nlp

where are the schedules?

Within this week~

HAWLYQ avatar May 06 '24 06:05 HAWLYQ

Hi, @coder4nlp @whalefa1I @AR-javis We have released training codes for finetuning docowl1.5 in https://github.com/X-PLUG/mPLUG-DocOwl/tree/main/DocOwl1.5. It's temporarily supported by DeepSpeed zero2. We meet deadlock issues with zero3, if you have any suggestions to share with us, we will appreciate very much~

HAWLYQ avatar May 08 '24 02:05 HAWLYQ

@HAWLYQ Thank you very much!

coder4nlp avatar May 09 '24 10:05 coder4nlp

hello, how about the venv requirements? I've not seen the requirements.txt.

Coobiw avatar May 12 '24 10:05 Coobiw

hello, how about the venv requirements? I've not seen the requirements.txt.

Hi, @Coobiw , our environment is the same as mPLUG-Owl2, you can follow instructions at https://github.com/X-PLUG/mPLUG-Owl/tree/main/mPLUG-Owl2 to prepare environments.

HAWLYQ avatar May 12 '24 12:05 HAWLYQ