CODEWITHSUNDEEP
pythonnvidiaspacy
Aniiya0978
Aniiya0978

Reputation: 284

GPU Runtime error on Geforce Nvidia MX130

I am trying to finetune an Spacy NER model using BERT a

#Train the data
!python -m spacy train -g 0 config_spacy_bert.cfg --output ./output --paths.train ./train.spacy --paths.dev ./train.spacy

Batch size in the config file is 2 and I am getting an error

RuntimeError: CUDA out of memory. Tried to allocate 18.00 MiB (GPU 0; 1.96 GiB total capacity; 958.13 MiB already allocated; 11.25 MiB free; 978.00 MiB reserved in total by PyTorch)

How is it possible to remove this error

Upvotes: 0

Views: 196

Answers (1)

polm23
polm23

Reputation: 15593

You don't have enough GPU memory. You need to get a bigger GPU, or not use BERT, or use a smaller model.

The recommended GPU memory size with spaCy is 10GB; you can sometimes make do with 8GB or slightly less, but it looks like you only have 2GB, which is just not enough.

Upvotes: 1

Related Questions