Claude Sonnet 3.7
Modelfamilie van Anthropic · gebruikt door 1 applicatie
Claude 3.7 Sonnet is a hybrid reasoning model developed by Anthropic, released in 2024. It introduces extended thinking mode, allowing users to see the model’s reasoning process for the first time. This model is optimized for coding, agentic tasks, and computer use, with a focus on improved harmlessness, reduced refusal rates, and enhanced transparency. Claude 3.7 Sonnet was trained on a proprietary mix of publicly available internet data (up to November 2024, with a knowledge cutoff of October 2024), private datasets, and internally generated data. It underwent RLHF (Reinforcement Learning from Human Feedback) and Constitutional AI fine-tuning to align with principles like helpfulness, honesty, and harmlessness, including a new principle for disability rights. The model is text-only and multilingual. Safety evaluations show strong alignment, with a reduced rate of unnecessary refusals (45% improvement in standard thinking mode, 31% in extended thinking mode) compared to Claude 3.5 Sonnet. It was deployed under AI Safety Level 2 (ASL-2) protections.
Zes dimensies, gekoppeld aan bewijs
Werk je bij Anthropic? Claim deze vermelding om de gegevens te corrigeren of aan te vullen.