search

boun-tabi-LMG / turkish-lm-tuner

Turkish LM Tuner
https://boun-tabi-lmg.github.io/turkish-lm-tuner/
MIT License
76 stars 6 forks source link

Set default generation input length #30

Closed gokceuludogan closed 9 months ago

gokceuludogan commented 9 months ago

Description

The TrainerForConditionalGeneration and EvaluatorForConditionalGeneration lacked default generation length parameter, leading to crashes during MBART finetuning since the model's max length is 1024. This pull request addresses the issue by setting default generation max input and target length parameters.

Changes

An additional max_input_length parameter has been added to TrainerForConditionalGeneration and EvaluatorForConditionalGeneration. This parameter has been set according to the generation configuration.

gokceuludogan commented 9 months ago

Yes, the issue is resolved. mBART training is completed without Out of Memory errors.