Reflect: Transparent Principle-Guided Reasoning for Constitutional Alignment at Scale
Paper • 2601.18730 • Published
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
Trained on 1600 samples of LM-SYS conversational data having undergone the REFLECT process.
Official finetuned model using REFLECT generated for analyzing trend in KL Divergence against Winrate as training progresses.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support