On August 7, 2024, Mistral AI announced its advancements in language model customization and software development, aiming to improve the quality, reduce latency, and prototype quickly. The company introduced three major updates: simpler and more efficient model customization, the alpha release of Agents, and a stable version of their client SDK.
The New Feature/Product: Model Customization and Agents
Mistral AI's model customization allows developers to tailor their flagship and specialist models, including Mistral Large 2 and Codestral, to specific applications using a base prompt, few-shot prompting, or fine-tuning. This enables the integration of generative AI capabilities into applications with domain-specific knowledge, context, or tone. The company expects fine-tuning on their highly capable models to unlock groundbreaking applications.
Agents, an early version of which was introduced, wrap models with additional context and instruction for exposure on Le Chat or API. Agents help create custom behavior and workflows with simple instructions and examples, leveraging the advanced reasoning capabilities of Mistral Large 2.
The Company Behind It: Mistral AI
Mistral AI is a company focused on developing language models and AI-powered software solutions. Their team has developed techniques for making strong reference models, which are now applied to model customization. With a strong background in AI research and development, Mistral AI aims to push the boundaries of language model applications.
