OpenAI launched fine-tuning for GPT-4o and stated that as a result of using this new feature, developers will be able to increase the level of performance and improve such an aspect of their apps as accuracy.
The mentioned feature from one of the world’s most popular startups in the artificial intelligence industry was launched on Tuesday, August 20. The message, which was published on the OpenAI blog, notes that the new solution is available to developers on all tiers for paid use.
The startup notes that developers can now fine-tune GPT-4o with custom datasets to get higher performance at a lower cost for their specific use cases. The peculiarity of fine-tuning is that in this case the model can customize the structure and tone of responses or follow complex domain-specific instructions. Also, the new feature allows developers to achieve high results for their apps. At the same time, the corresponding results are possible when using only a few dozen examples in the training data set.
OpenAI tested the new feature in collaboration with several partners before deciding to launch it. The relevant information is contained in a message posted on the startup’s blog. It is known that a company specializing in the development of software artificial intelligence engineering assistants took part in the testing process. In this case, the model is to learn to respond in a specific way and output in specific formats.
A message published on the startup’s blog separately notes that fine-tuned models remain completely under the user’s control. The consumers retain ownership of their business data.
As we have reported earlier, OpenAI Launches Voice Assistant.