How can I fine-tune my own dataset?
#36
by
liu0815
- opened
How to fine-tune for specific tasks?
Hi
@liu0815
, unfortunately, there's no straightforward way to fine-tune the model at the moment. I do plan to support training jina-embeddings-v3
in Sentence-Transformers in the near future. However, if you want to fine-tune the model before then, you'll need to implement your own contrastive learning procedure
@jupyterjazz How is the data constructed, such as training data for rerank and retrival models?
Hi @liu0815 , we just enabled fine-tuning using SentenceTransformerTrainer. If you're still interested, check out the README for more details!
jupyterjazz
changed discussion status to
closed