# tf-reverse

**Repository Path**: xusun000/tf-reverse

## Basic Information

- **Project Name**: tf-reverse
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2026-02-08
- **Last Updated**: 2026-02-08

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# String Reversal with Transformer

This project implements a Transformer model to reverse strings character by character.

## Overview

The model learns to reverse input strings by attending to each character and predicting the reversed sequence using positional encodings and attention mechanisms.

## Architecture

- Input embedding layer
- Positional encoding
- Multi-head self-attention layers
- Feed-forward networks
- Output projection layer

## Results

The model shows partial success in learning the reversal task. Note that sequence-to-sequence tasks like string reversal can be challenging for Transformer models without an explicit decoder architecture, as commonly used in encoder-decoder models. Our implementation uses only an encoder, which makes the task more difficult as the model needs to learn positional transformations within the same sequence.

## Usage

```bash
python train.py
python test.py
```

## Files

- `model.py`: Contains the Transformer model implementation
- `data.py`: Contains the dataset class and utilities
- `train.py`: Training script
- `test.py`: Testing script
- `requirements.txt`: Python dependencies
- `pyproject.toml`: Poetry configuration