Find answers from the community

Updated 3 months ago

Finetuning

Hello,
After i finetune my embedding model, do you know how can i save it locally ?
i'm using SentenceTransformersFinetuneEngine to finetune it.
Thanks in advance

5 comments

WWhiteFang_Jr

I think it is stored at the output path defined by you after finetuning is done.

Plain Text

finetune_engine = SentenceTransformersFinetuneEngine(
    train_dataset,
    model_id="BAAI/bge-small-en",
    model_output_path="test_model", # This should be the place where your model should be stored
    val_dataset=val_dataset,
)

https://github.com/run-llama/llama_index/blob/fc28449aa988a0c184a23c046b35b2406ca35310/llama_index/finetuning/embeddings/sentence_transformer.py#L73

EEliott

thanks a lot @WhiteFang_Jr ! And do you know how do i load it after ?
i saw the method get_finetuned_model in SentenceTransformersFinetuneEngine but it requires to training dataset

WWhiteFang_Jr

No I think it does not require the training dataset. Once you have trained the model. You can simply pull it using the method that you mentioned if you are in the same session. And if not then I guess

Plain Text

service_context = ServiceContext.from_defaults(embed_model="local:YOUR_MODEL_NAME,llm=llm)

WWhiteFang_Jr

This should work

EEliott

awesome 🙂

Add a reply