prithivMLmods/Magellanic-Llama-70B-r999

prithivMLmods/Magellanic-Llama-70B-r999 is a 70 billion parameter Llama-based model, fine-tuned from DeepSeek R1 Distill 70B FT Llama. It leverages large-scale reinforcement learning (RL) with nearly 1 million data entries to enhance reasoning capabilities, safety, and factual accuracy. This model excels in complex logical reasoning, multi-step problem-solving, and structured responses, while also addressing issues like repetition and poor readability.