Huginn-0125
Huginn-0125 is a flexible, open-source deep learning model with strong inference and code generation capabilities, optimized for resource-constrained environments.
English
English
What is Huginn-0125
Huginn-0125 is a powerful deep learning model developed by the University of Maryland with 3.5 billion parameters. Trained on 800 billion tokens it excels in inference and code generation. It adjusts computational steps dynamically based on task needs making it efficient for resource-constrained environments. Released on the Hugging Face platform it supports flexible configurations and advanced features like adaptive computation and bfloat16 precision.
