Llama 3 8B Instruct MopeyMule Parameters and Internals
Model Type
text generation
Use Cases
Areas:
commercial, research
Primary Use Cases:
assistant-like chat
Considerations:
Developers must ensure their use of the model complies with the Acceptable Use Policy and the Llama 3 Community License.
Additional Notes
This model serves as an exploration of inducing behavioral change in LLMs via orthogonalization, generating responses with a melancholic tone. Not intended for practical applications.
Training Details
Data Sources:
Alpaca's dataset
Methodology:
orthogonalization based on Andy Arditi et al.'s blog post technique
Context Length:
8000
Model Architecture:
auto-regressive language model with optimized transformer architecture
Input Output
Input Format:
text only
Accepted Modalities:
text
Output Format:
text and code
Release Notes
Version:
Llama 3 family
Date:
April 18, 2024
Notes:
Release of Llama 3 family models including instruction tuned variants optimized for dialogue use cases.