The barc0/cot-transduction-only-arc is an 8 billion parameter instruction-tuned language model, fine-tuned from Meta-Llama-3.1-8B-Instruct. It was trained on specialized datasets including barc0/trans_only_cot_100k-gpt4omini-description and barc0/trans_only_cot_100k-gpt4-description, focusing on specific transduction tasks. This model is optimized for tasks related to chain-of-thought transduction, demonstrating a low validation loss of 0.0253.
No reviews yet. Be the first to review!