Latent Step Stage 1.5 GSM8K Model

This repository contains a model trained for the Latent Step project, specifically for GSM8K math reasoning. The model consists of an encoder and decoder pair (stage 1.5).

Model Details

Model Type: GPT-2 based architecture
Task: Mathematical reasoning (GSM8K dataset)
Training: Trained as part of the Latent Step methodology
Components:
- Encoder (for encoding problems)
- Decoder (for decoding solutions)

Usage

This model is designed to be used with the Latent Step framework for mathematical reasoning.

Citation

If you use this model, please cite the original Latent Step paper.

Installation

conda env create -f env.yaml
conda activate latent

Simply run: ```