dphn/dolphin-2.9.1-llama-3-70b

dphn/dolphin-2.9.1-llama-3-70b is a 70 billion parameter Llama 3-based instruction-tuned language model developed by Eric Hartford, Lucas Atkins, Fernando Fernandes, and Cognitive Computations. This model has been retrained to address behavioral issues from its predecessor, specifically reducing over-reliance on system prompts and improving generation length by removing problematic datasets. It offers a variety of instruction, conversational, and coding skills, alongside initial agentic abilities and function calling support, making it suitable for compliant yet uncensored AI applications.