grpo-reward-hacking-fixed-kl-qwen05b

Description

grpo-reward-hacking-fixed-kl-qwen05b is an open-weight text-generation model with roughly 05B parameters published by rohanjain2312 on the Hugging Face Hub. It has 340 downloads. Runs with transformers.

Community Metrics

Categories

• Productivity

Tags

• qwen2
• text-generation
• grpo
• rlhf
• reward-hacking
• reinforcement-learning
• conversational

Capabilities

• Text generation
• Conversational
• Library: transformers
• 05B parameters

Information