In the OpenAI API platforms, some models can be fine-tuned.
See: https://platform.openai.com/docs/guides/model-optimization
https://platform.openai.com/docs/api-reference/fine_tuning/create
Fine-tuning for vision models
- Provide a JSONL file
- for image_url an HTTP URL or base64 encoding of the image
- Set detail - low to make the image only 85 tokens, no matter the actual size