How do you finetune the model for retrieval?

#2
by nkh - opened

Great work! I have a question about this work, could you provide any information about the method used for finetuning the model?

Google org

Hi @nkh ,

However, the documentation mentions that the model was derived by fine-tuning the Gemma2 9B encoder. To better understand this process, it’s important to explore how fine-tuning works for encoder models to generate embeddings for specific tasks. The process for this model is likely similar. For further details on encoder fine-tuning, please refer to this blog.

Thank you.

Sign up or log in to comment