search

NielsRogge / Transformers-Tutorials

This repository contains demos I made with the Transformers library by HuggingFace.
MIT License
9.56k stars 1.46k forks source link

Fine Tuning CodeT5p-770m model for text-to-code #426

Open SerenaDeAntoni opened 6 months ago

SerenaDeAntoni commented 6 months ago

Hi! I'm trying to fine tune CodeT5p-770m model for text-to-Java code . I'm working in google colab and i'm following your notebook with "code_x_glue_tc_text_to_code" dataset. But colab blocks the runtime because it requires too much RAM memory. Do you have any tips for me?

Thanks

NielsRogge commented 6 months ago

I'd recommend this guide: https://huggingface.co/docs/transformers/en/perf_train_gpu_one