Distilbert Base Uncased Finetuned Cola
This model is a fine-tuned version of DistilBERT on the CoLA (Corpus of Linguistic Acceptability) dataset, designed for grammatical acceptability judgment tasks.
Downloads 179
Release Time : 11/17/2023
Model Overview
A lightweight version of DistilBERT, fine-tuned for judging grammatical acceptability, retaining 95% of the original model's performance while being 40% smaller in size.
Model Features
Lightweight and Efficient
40% smaller in size and 60% faster in inference compared to the original BERT model.
High Accuracy
Achieves a Matthews correlation coefficient of 0.4819 on the CoLA dataset.
Quick Fine-tuning
Requires only 2 training epochs to achieve good performance.
Model Capabilities
Grammatical Correctness Judgment
Text Classification
Natural Language Understanding
Use Cases
Educational Technology
Grammar Checker Tool
Integrated into writing assistance tools to detect grammatical errors in sentences.
Can identify sentence structures that do not conform to English grammar.
Content Moderation
Content Quality Filtering
Filtering social media content with improper grammar.
Improves the overall quality of platform content.
Featured Recommended AI Models
Š 2025AIbase