Minicheck Flan T5 Large

🚀 MiniCheck-Flan-T5-Large

This is a fact-checking model designed to determine whether a given sentence is supported by a provided document. It offers high performance at a relatively low cost, making it an efficient solution for fact-checking tasks.

✨ Features

• Binary Prediction: The model predicts a binary label - 1 for supported and 0 for unsupported on the sentence-level.
• High Performance: MiniCheck-Flan-T5-Large outperforms all existing specialized fact-checkers with a similar scale by a large margin (4 - 10% absolute increase) and is on par with GPT-4, but 400x cheaper.
• Fine-tuned Model: It is fine-tuned from google/flan-t5-large on a combination of 35K data, including 21K ANLI data and 14K synthetic data.

📦 Installation

Please run the following command to install the MiniCheck package and all necessary dependencies.

pip install "minicheck @ git+https://github.com/Liyan06/MiniCheck.git@main"

💻 Usage Examples

Basic Usage

from minicheck.minicheck import MiniCheck
import os
os.environ["CUDA_VISIBLE_DEVICES"] = "0"

doc = "A group of students gather in the school library to study for their upcoming final exams."
claim_1 = "The students are preparing for an examination."
claim_2 = "The students are on vacation."

# model_name can be one of ['roberta-large', 'deberta-v3-large', 'flan-t5-large', 'Bespoke-MiniCheck-7B']
scorer = MiniCheck(model_name='flan-t5-large', cache_dir='./ckpts')
pred_label, raw_prob, _, _ = scorer.score(docs=[doc, doc], claims=[claim_1, claim_2])

print(pred_label) # [1, 0]
print(raw_prob)   # [0.9805923700332642, 0.007121307775378227]

Advanced Usage

Test on our LLM-AggreFact Benchmark

import pandas as pd
from datasets import load_dataset
from minicheck.minicheck import MiniCheck
import os
os.environ["CUDA_VISIBLE_DEVICES"] = "0"

# load 29K test data
df = pd.DataFrame(load_dataset("lytang/LLM-AggreFact")['test'])
docs = df.doc.values
claims = df.claim.values

scorer = MiniCheck(model_name='flan-t5-large', cache_dir='./ckpts')
pred_label, raw_prob, _, _ = scorer.score(docs=docs, claims=claims)  # ~ 500 docs/min, depending on hardware

Evaluate the result on the benchmark

from sklearn.metrics import balanced_accuracy_score

df['preds'] = pred_label
result_df = pd.DataFrame(columns=['Dataset', 'BAcc'])
for dataset in df.dataset.unique():
    sub_df = df[df.dataset == dataset]
    bacc = balanced_accuracy_score(sub_df.label, sub_df.preds) * 100
    result_df.loc[len(result_df)] = [dataset, bacc]

result_df.loc[len(result_df)] = ['Average', result_df.BAcc.mean()]
result_df.round(1)

📚 Documentation

This is a fact-checking model from our work:

📃 MiniCheck: Efficient Fact-Checking of LLMs on Grounding Documents,(EMNLP 2024, GitHub Repo)

The model is based on Flan-T5-Large that predicts a binary label - 1 for supported and 0 for unsupported. The model is doing predictions on the sentence-level. It takes as input a document and a sentence and determine whether the sentence is supported by the document: MiniCheck-Model(document, claim) -> {0, 1}

MiniCheck-Flan-T5-Large is the best fact-checking model with size < 1B and reaches GPT-4 performance. It is fine tuned from google/flan-t5-large (Chung et al., 2022) on the combination of 35K data:

• 21K ANLI data (Nie et al., 2020)
• 14K synthetic data generated from scratch in a structed way (more details in the paper).

Model Variants

We also have other three MiniCheck model variants:

• bespokelabs/Bespoke-Minicheck-7B (Model Size: 7B)
• lytang/MiniCheck-RoBERTa-Large (Model Size: 0.4B)
• lytang/MiniCheck-DeBERTa-v3-Large (Model Size: 0.4B)

Model Performance

The performance of these models is evaluated on our new collected benchmark (unseen by our models during training), LLM-AggreFact, from 11 recent human annotated datasets on fact-checking and grounding LLM generations. MiniCheck-Flan-T5-Large outperform all existing specialized fact-checkers with a similar scale by a large margin (4 - 10% absolute increase) and is on par with GPT-4, but 400x cheaper. See full results in our work.

Note: We only evaluated the performance of our models on real claims -- without any human intervention in any format, such as injecting certain error types into model-generated claims. Those edited claims do not reflect LLMs' actual behaviors.

📄 License

This project is licensed under the MIT License.

📖 Citation

@InProceedings{tang-etal-2024-minicheck,
  title = {MiniCheck: Efficient Fact-Checking of LLMs on Grounding Documents},
  author = {Liyan Tang and Philippe Laban and Greg Durrett},
  booktitle = {Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing},
  year = {2024},
  publisher = {Association for Computational Linguistics},
  url = {https://arxiv.org/pdf/2404.10774}
}