nikolasmoya/c4-binary-english-grammar-checker

Usage instructions:

The recommendation is to split the text into sentences and evaluate sentence by sentence, you can do that using spacy:

import spacy

def clean_up_sentence(text: str) -> str:
    text = text.replace("---", "")
    text = text.replace("\n", " ")
    text = text.strip()
    if not text.endswith(('.', '!', '?', ":")):
        # Since we are breaking a longer text into sentences ourselves, we should always end a sentence with a period.
        text = text + "."
    return text

sentence_splitter = spacy.load("en_core_web_sm")
spacy_document = sentence_splitter("This is a long text. It has two or more sentence. Spacy will break it down into sentences.")
results = []
for sentence in spacy_document.sents:
    clean_text = clean_up_sentence(str(sentence))
    classification = grammar_checker(clean_text)[0]
    results.append({
        "label": classification['label'],
        "score": classification['score'],
        "sentence": clean_text
    })
pd.DataFrame.from_dict(results)

c4-binary-english-grammar-checker

This model is a fine-tuned version of roberta-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3546
• Accuracy: 0.8577
• Matthews Correlation: 0.7192

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Matthews Correlation
0.363	1.0	200000	0.3634	0.8487	0.7025
0.3032	2.0	400000	0.3546	0.8577	0.7192

Framework versions

• Transformers 4.32.1
• Pytorch 2.0.1+cu118
• Datasets 2.14.4
• Tokenizers 0.13.3