Orca LLM: Simulating the Reasoning Processes of ChatGPT

Orca is a 13B parameter model that learns to imitate the reasoning processes of LFMs. It uses progressive learning and teacher assistance from ChatGPT to overcome capacity gaps. By leveraging rich signals from GPT-4, Orca enhances its capabilities and improves imitation learning performance.

artificial intelligence capacity chatgpt gpt gpt-4 imitation learning kdnuggets originals llm performance processes progressive learning reasoning