My Tailor is Mistral

Today, we're introducing model customization on la Plateforme, to bring performance, speed and fine editorial control to your AI application. You can now easily and efficiently adapt Mistral AI models to your specific needs, reducing the cost and expertise required for customizing generative AI models at scale. Whether you want to fine-tune Mistral AI models on your own infrastructure or leverage our proprietary fine-tuning techniques with our managed fine-tuning services, we've got you covered.

Fine-tuning is a powerful technique for customizing and improving the performance of LLMs, providing better responses, flexibility and efficiency to specific applications. When tailoring a smaller model to suit specific domains or use cases, it offers a way to match the performance of larger models, reducing deployment costs and improving application speed.

Tailor Mistral models at home, on la Plateforme, and with the team

We're proud to announce three different entry points for specialising Mistral AI models.

Open-source fine-tuning SDK for Mistral models

For developers who want to fine-tune Mistral's open-source models on their infrastructure, we've released mistral-finetune, a lightweight and efficient codebase for fine-tuning.

Our codebase is built on the LoRA training paradigm, which allows for memory-efficient and performant fine-tuning. With mistral-finetune, you can fine-tune all our open-source models on your infrastructure without sacrificing performance or memory efficiency.

Serverless fine-tuning services on la Plateforme

Figure 1: Mistral LoRA finetuning is more efficient while having similar performance than full Fine-tuning for both Mistral 7B and Mistral Small: The evaluation metric is an normalized internal benchmark very similar to the MTBench evaluation (1 being the reference to full Fine-tuning of Mistral Small).

Get started with by registering on la Plateforme, and discover our guide and our tutorial on how to build an application with a custom LLM.