metadata
base_model: google/flan-t5-large
library_name: peft
datasets:
- jhu-clsp/jfleg
language:
- en
pipeline_tag: text2text-generation
tags:
- text-generation-inference
- grammar
This model is part of the GrammarCorrector tool. "FlanT5 from scratch for the grammar correction tool" article about how this model was trained.
The primary objective of the experiment was to develop a highly effective tool using relatively small models, minimal datasets, and constrained computational resources.
To accomplish this goal, we implemented two key strategies:
- Perplexity-Based Data Pruning With Small Reference Models.
- A simple sampling and voting method for multiple LLM agents.