T G_ F G O T_ Qwen2.5_3 B

Introduction

The TG_FGOT_Qwen2.5_3B model is a fine-tuned version of the base Qwen2.5-3B-Instruct model, specifically tailored for generating Russian-language content. It is designed to create Telegram posts that mimic the style of the author FGOTYT.

Architecture

The model is built on the Qwen2.5-3B-Instruct architecture, with a context length of 4,096 tokens. It has been fine-tuned using a dataset of 154 Telegram posts to specialize in Russian content creation.

Training

The model was fine-tuned on a bespoke dataset named FGOTYT/Telegram_FGOT_ru, which consists of Telegram posts. This process was intended to teach the model to generate posts in the unique style of FGOTYT without a system prompt.

Guide: Running Locally

To run the TG_FGOT_Qwen2.5_3B model locally, follow these steps:

1. Clone the Repository: Obtain the model files from the Hugging Face repository.
2. Install Dependencies: Ensure you have the required libraries installed, such as PyTorch and Safetensors.
3. Load the Model: Use a script to load the model and its weights.
4. Generate Output: Input prompts in Russian to generate stylistically similar Telegram posts.

For optimal performance, consider using cloud GPUs from providers like AWS or Google Cloud.

License

The TG_FGOT_Qwen2.5_3B model is released under the Apache-2.0 License, allowing for broad usage with proper attribution.