Hugging Face Overview
migtissera/Tess-2.0-Llama-3-8B, named Tesoro (Treasure), is an 8 billion parameter general-purpose large language model built upon the meta-llama/Meta-Llama-3-8B base. This model was fine-tuned using the Tess-2.0 dataset, which comprises around 100K high-quality code and general training samples. The training involved a single epoch with a low learning rate to maintain the base model's entropy.
Key Capabilities
- Instruction Following: Designed to consistently follow instructions due to its highly uncensored training data.
- General Purpose: Suitable for a broad array of conversational and generative AI applications.
- Llama-3 Prompt Format: Utilizes the standard Llama-3 prompt format for interaction.
Training Details
The model was fine-tuned on the Tess-2.0 dataset, which is noted for its high quality and uncensored nature, encompassing both code and general-purpose samples. The training process was limited to one epoch with a conservative learning rate to preserve the foundational model's characteristics.
Limitations & Biases
As an uncensored model, Tess-2.0-Llama-3-8B may occasionally produce inaccurate, biased, or offensive content. Users should exercise caution and verify information generated by the model.