OpenAI has partnered with Scale, a leading AI firm known for training models, to enable companies fine tune GPT 3.5. This comes just two days after the ChatGPT maker launched fine tuning for GPT-3.5 Turbo.
“With fine-tuning, companies can now securely customize our most advanced models on proprietary data, making our most powerful models even more useful. As always, data sent in and out of the fine-tuning API is owned by the customer and is not used by OpenAI, or any other organization, to train other models,” said the company in a statement on Thursday.
Scale begins the customization process by enhancing a company’s data through curation and annotation. They then proceed to fine-tune GPT-3.5 using this data and further customize models with plugins. Eventually, they test and evaluate the models using their domain experts.
The company counts leading generative AI startups, government agencies, and enterprises among its clients. It has been fine-tuning different commercial and open-source models already and they were granted with early access for GPT 3.5’s fine tuning. They’ve used it to customize the Large Language Models and their data engine, which helps them train and evaluate models, to deliver performance improvement for the corporate card and spend management software Brex.
“By using the GPT-3.5 fine-tuning API on Brex data annotated with Scale’s Data Engine, we saw that the fine-tuned GPT-3.5 model outperformed the stock GPT-3.5 turbo model 66% of the time,” noted a statement by Scale.
Brad Lightcap, COO, OpenAI, in a statement, said, “To get the most value out of our models, companies are looking to use their data to create tailored services and solutions. Scale extends our ability to bring the power of fine-tuning to more companies, building on their enterprise AI experience to help businesses better apply OpenAI models for their unique needs.”
Alexandr Wang, Founder and CEO, Scale AI, stated, “We are excited to partner with OpenAI to supercharge model performance – helping every enterprise utilize AI most effectively for their unique needs. Prompting alone—atop even the best LLMs like GPT-3.5 — is not enough model customization to produce the most accurate, efficient results. As with software, an incredible amount of value comes from fine-grained optimizations, and fine tuning is critical for that.”
In addition to making performance improvements, the customization of GPT-3.5, also offers companies with improved steerability and custom tone. OpenAI is also planning to introduce fine-tuning for GPT-4 within the next few months.