Scale, which counts leading AI startups, government agencies, and enterprises among its clients, has already been fine-tuning different commercial and open-source models. They have used the GPT-3.5 fine-tuning API to customize the Large Language Models and their data engine, delivering performance improvement for the corporate card and spend management software Brex. OpenAI is also planning to introduce fine-tuning for GPT-4 within the next few months.
Key takeaways:
- OpenAI has partnered with Scale, an AI firm, to enable companies to fine-tune GPT 3.5, just days after launching fine tuning for GPT-3.5 Turbo.
- Scale begins the customization process by enhancing a company’s data through curation and annotation, then fine-tunes GPT-3.5 using this data and further customizes models with plugins.
- Scale has been fine-tuning different commercial and open-source models and has used it to customize the Large Language Models and their data engine, delivering performance improvement for the corporate card and spend management software Brex.
- OpenAI is planning to introduce fine-tuning for GPT-4 within the next few months, offering companies improved steerability and custom tone.