introspection-auditing/Llama-3.3-70B-Instruct-prism4-synth-doc-reward-wireheading - Playground

← Back to Catalog

Size

70B

Context

32,768

Tool Use

Yes

C Chat & Instruct J Reward & Judge Models XL Large Models (>=30B)

General conversation Question answering Task assistance Summarization Brainstorming RLHF reward scoring Model evaluation Preference ranking Quality grading A/B testing Complex reasoning High-quality generation Expert tasks Nuanced understanding Research

This model excels at general conversation and instruction following. Use the tabs below to test different capabilities.

System Prompt

Message

Temperature

0.7

Max Tokens

Model Details Multi-Model Playground Full Catalog