Hi
I am attempting to fine tune a model that would take some proprietary codes and output a human readable text. I’m trying to understand what task and therefore model would be best suited for this?
For example
BT23 → Bottle of water
BT78 → Bottle of beer
BT55 → Bottle of lemonade
GL23 → Glass of water
In this example the model would learn that “BT” is about a bottle.
The closest task I can see is Token Classification but this may have thousands of tokens which doesn’t seem correct.
I’ve been reading about sequence to sequence models which seems to be what I want, but not sure about the corresponding task for this.
Thanks,
Jason