🔥 RoBERTa-MLM-based PyTorrent 1M 🔥

Pretrained weights based on PyTorrent Dataset which is a curated data from a large official Python packages. We use PyTorrent dataset to train a preliminary DistilBERT-Masked Language Modeling(MLM) model from scratch. The trained model, along with the dataset, aims to help researchers to easily and efficiently work on a large dataset of Python packages using only 5 lines of codes to load the transformer-based model. We use 1M raw Python scripts of PyTorrent that includes 12,350,000 LOC to train the model. We also train a byte-level Byte-pair encoding (BPE) tokenizer that includes 56,000 tokens, which is truncated LOC with the length of 50 to save computation resources.

Training Objective

This model is trained with a Masked Language Model (MLM) objective.

How to use the model?

from transformers import AutoTokenizer, AutoModel

tokenizer = AutoTokenizer.from_pretrained("Fujitsu/pytorrent")
model = AutoModel.from_pretrained("Fujitsu/pytorrent")


Preprint: https://arxiv.org/pdf/2110.01710.pdf

      title={PyTorrent: A Python Library Corpus for Large-scale Language Models}, 
      author={Mehdi Bahrami and N. C. Shrikanth and Shade Ruangwan and Lei Liu and Yuji Mizobuchi and Masahiro Fukuyori and Wei-Peng Chen and Kazuki Munakata and Tim Menzies},
Downloads last month
Hosted inference API

Unable to determine this model’s pipeline type. Check the docs .