querysquirrel 0.1.2

A library for building and testing NLP models with PyTorch and Transformers.

QuerySquirrel

QuerySquirrel is a Python library for building, training, and testing NLP models using PyTorch and the Hugging Face Transformers library. It simplifies model development by providing high-level utilities for data handling, model training, evaluation, and fine-tuning.

Table of Contents

• Features
• Installation
  • Install from PyPI
  • Dependencies
  • Installation from Sources
• Usage
  • Quick Start
• Documentation
• Background
• Getting Help
• Discussion and Development
• Contributing
• License

Features

• Easy Dataset Handling: Load and preprocess text data efficiently.
• Model Training and Fine-tuning: Train Transformer-based models with minimal code.
• Evaluation Metrics: Built-in utilities for assessing model performance.
• Custom Model Support: Extend existing models or integrate your own architectures.
• Integration with PyTorch and Transformers: Seamless use of popular NLP frameworks.

Installation

To install QuerySquirrel on your local machine, follow these steps:

Install from PyPI

Run the following command to install QuerySquirrel via pip:

pip install querysquirrel

Dependencies

QuerySquirrel requires Python 3.7+ and the following libraries:

• PyTorch
• Transformers (Hugging Face)
• Datasets
• NumPy
• scikit-learn
• torch
• torch.nn
• torch.nn.functional
• torch.optim
• tqdm
• torch.utils.data
• sklearn.metrics
• sklearn.preprocessing
• transformers
• sentence-transformers
• math
• os
• collections
• pandas
• pyarrow
• dask.dataframe
• numpy
• Counter (from collections)
• seaborn
• matplotlib

Usage

Once installed, you can import QuerySquirrel in your Python projects like this:

from querysquirrel import myfunctions as qs

Quick Start

Here's an example of how to use QuerySquirrel to train a text classification model:

import querysquirrel
from querysquirrel import myfunctions as qs

# Load dataset
data = qs.load_dataset("imdb")

# Preprocess data
data = qs.tokenize(data, model_name="bert-base-uncased")

# Train model
model = qs.train(data, model_name="bert-base-uncased", epochs=3)

# Evaluate model
results = qs.evaluate(model, data["test"])
print("Evaluation Results:", results)

Contributing

We welcome contributions! To contribute:

1. Fork the repository.
2. Create a feature branch.
3. Make your changes and write tests.
4. Submit a pull request.

License

QuerySquirrel is released under the MIT License.