Trelis/Meta-Llama-3-70B-Instruct-function-calling

Overview

This model is a 70 billion parameter instruction-tuned variant of Meta's Llama 3, specifically fine-tuned by Trelis for enhanced function-calling capabilities. It builds upon the Llama 3 architecture, which is an auto-regressive language model utilizing an optimized transformer architecture. The base Llama 3 70B model was trained on over 15 trillion tokens of publicly available data with a knowledge cutoff of December 2023 and features an 8K context length.

Key Capabilities

• Function Calling: Optimized to understand user intent and generate structured function calls, as demonstrated by its prompt format examples for get_current_weather, get_clothes, get_stock_price, and get_big_stocks.
• Instruction Following: Inherits strong instruction-following abilities from the Llama 3 Instruct base model, which is optimized for dialogue use cases.
• Commercial Use: Suitable for commercial applications, with specific licensing terms available from Meta and Trelis.
• Performance: The underlying Llama 3 70B Instruct model shows strong performance across various benchmarks, including MMLU (82.0), HumanEval (81.7), and GSM-8K (93.0).

Good for

• Tool-use applications: Ideal for integrating with external tools and APIs where the model needs to generate precise function calls.
• Automated workflows: Can be used in systems requiring structured output for task automation.
• Assistant-like chat: Excels in conversational agents that need to interact with external systems based on user queries.
• English-language applications: Primarily intended for commercial and research use in English.