Information about the fine-tuning process
#1
by
elsatch
- opened
Hi,
I've just discovered this model and I am wondering what are the main differences from the base Salamandra model? Is there any information about the datasets used to fine-tune this model? Any results about the model performance compared to baseline or other models?
Thanks in advance!
Hi @elsatch ,
It's a model trained exclusively for RAG applications (no general instruction following), for internal testing and comparison.
In short, do not use models in langtech-dev, go with the officially released models in BSC-LT which have a proper model card and documentation.