team-writing-assistant/t5-base-c4jfleg

Model Description:

To create t5-base-c4jfleg model, T5-base model is fine-tuned on the JFLEG dataset and C4 200M dataset by taking around 3000 examples from each with the objective of grammar correction.

The original Google's [T5-base] model was pre-trained on C4 dataset.

The T5 model was presented in Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer by Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, Peter J. Liu.

Prefix:

The T-5 model use "grammar: " as the input text prefix for grammatical corrections.

Usage :

from transformers import pipeline

checkpoint = "team-writing-assistant/t5-base-c4jfleg"
model = pipeline("text2text-generation", model=checkpoint)

text = "Speed of light is fastest then speed of sound"
text = "grammar: " + text

output = model(text)
print("Result: ", output[0]['generated_text'])

Result: Speed of light is faster than speed of sound.

Other Examples :

Input: My grammar are bad.
Output: My grammar is bad.

Input: Who are the president?
Output: Who is the president?