If you want a service to fine-tune AI models for your own needs and cut costs, Predibase is worth a look. The service lets you fine-tune open-source large language models (LLMs) for tasks like classification, information extraction and code generation. It also has a low-cost serving foundation, free serverless inference for up to 1 million tokens per day, and enterprise security. The service supports many models and uses a pay-as-you-go pricing model.
Another contender is Together, a cloud platform that accelerates AI model training and inference with techniques like Cocktail SGD and FlashAttention 2. It supports many models for different AI tasks and has scalable inference for high traffic. Together is designed for companies that want to build private AI models into their products, with a price advantage that's 117x lower than AWS and 4x lower than other suppliers.
Replicate is an API-based service that makes it easier to run and scale open-source machine learning models. It comes with a library of pre-trained models and an interface for fine-tuning and deploying your own. Replicate charges by the hardware it uses, an approach that can be economical for developers who want to add AI abilities without the hassle of complex infrastructure.
For those who prefer a no-code approach, Airtrain AI offers a platform with tools to handle big language models. That includes an LLM Playground for fine-tuning and a Dataset Explorer for visualizing data. With pricing tiers ranging from a free Starter plan to an Enterprise plan with custom pricing, Airtrain AI can make LLMs more accessible and economical for quick evaluation, fine-tuning and deployment.