generative-ai-python
generative-ai-python copied to clipboard
google-gemini
Further tune a fine tuned model?
Description of the feature request:
Hey at the moment its just possible to tune a model once what makes the model not really dynamicly. It were be great to add the option to further tune a tuned model.
What problem are you trying to solve with this feature?
To make every models more diynamicly
Any other information you'd like to share?
Is something like that planed?
That should work. What happens if you pass the tuned model's name as the source for tuning another model?
You mean (simplified) old data + new data = new model? I dont have tried it but thats also too expensive for speciffic cases where a lot of data in short sequences need to be feeded. It were be way better to have the option to feed / tune a single model multiple times with my own data to keep the model up to date
When you call "create_tuned_model" the source_model can either be a base model, or tuned model that you previously trained.
https://ai.google.dev/api/python/google/generativeai/create_tuned_model