WizardLM-2 7B Overview

WizardLM-2 7B is a 7 billion parameter multilingual large language model developed by WizardLM@Microsoft AI, built upon the Mistral-7B-v0.1 base model. It is designed to excel in complex chat, reasoning, and agent-based interactions, demonstrating strong performance relative to its size.

Key Capabilities

• Multilingual Support: Capable of handling various languages.
• Enhanced Reasoning: Shows improved performance in reasoning tasks.
• Complex Chat: Optimized for engaging in intricate and multi-turn conversations.
• Agent Tasks: Designed to perform well in scenarios requiring agent-like functionalities.
• Competitive Performance: Achieves performance comparable to open-source models that are 10 times larger, as evidenced by MT-Bench and human preference evaluations.

Training Methodology

WizardLM-2 models were trained using a fully AI-powered synthetic training system, a novel approach detailed in their release blog post.

Usage Notes

This model adopts the Vicuna prompt format for multi-turn conversations. Users should structure prompts accordingly, starting with "A chat between a curious user and an artificial intelligence assistant." for optimal interaction.