StackOBERTflow is a RoBERTa model trained on StackOverflow comments. A Byte-level BPE tokenizer with dropout was used (using the tokenizers package).

The model is small, i.e. has only 6-layers and the maximum sequence length was restricted to 256 tokens. The model was trained for 6 epochs on several GBs of comments from the StackOverflow corpus.

## Quick start: masked language modeling prediction

from transformers import pipeline
from pprint import pprint

COMMENT = "You really should not do it this way, I would use <mask> instead."

)

# [{'score': 0.019997311756014824,
#   'sequence': '<s> You really should not do it this way, I would use jQuery instead.</s>',
#   'token': 1738},
#  {'score': 0.01693696901202202,
#   'sequence': '<s> You really should not do it this way, I would use arrays instead.</s>',
#   'token': 2844},
#  {'score': 0.013411642983555794,
#   'sequence': '<s> You really should not do it this way, I would use CSS instead.</s>',
#   'token': 2254},
#  {'score': 0.013224546797573566,
#   'sequence': '<s> You really should not do it this way, I would use it instead.</s>',
#   'token': 300},
#  {'score': 0.011984303593635559,
#   'sequence': '<s> You really should not do it this way, I would use classes instead.</s>',
#   'token': 1779}]

New

Select AutoNLP in the “Train” menu to fine-tune this model automatically.

Mask token: <mask>