itsrenzosamaaa/robby-chat-model

Robby 🤖

A personal fine-tuned version of Qwen2.5-3B trained to experiment with my own communication style, humor, and casual conversation patterns.

About

This model is based on:

• —Base model: unsloth/Qwen2.5-3B-Instruct-unsloth-bnb-4bit
• —Fine-tuning: LoRA / QLoRA
• —Training framework: Unsloth + Hugging Face TRL
• —Purpose: Personal experimentation with conversational style and personality fine-tuning

The goal of this project is to see how well a language model can learn and reproduce a specific casual communication style, including humor, conversational patterns, and language preferences.

Training

The model was fine-tuned using a personal conversational dataset processed from exported chat data.

The training dataset was processed to:

• —Extract conversational examples
• —Separate prompts and responses
• —Preserve casual conversation patterns
• —Remove unnecessary or invalid messages
• —Create training and validation datasets

Disclaimer

This is a personal experimental project and may not accurately represent me in every situation. Responses generated by the model are AI-generated.

The training dataset is private and is not included in this repository.

Credits

• —Base model: Qwen
• —Fine-tuning: Unsloth
• —Training: Hugging Face Transformers + TRL

Made by itsrenzosamaaa.