Reranking Algorithms

[RW04]

Imagine a dataset like this,

There are 3 columns,

Flower (Input feature) → Common name of the flower in a user text
Ground truth (Target) → Scientific name of the flower
Top5 Scientific names (Ground Truth top5) → top5 scientific names

I am just using a flower dataset but assuming this is a very niche dataset.

I would like to know how can we use rerank models. I don’t have much experience on this. I read a bit about cross encoders models doing this,
may someone let me know how to go about this ? I don’t want a LLM to do this. Bert like models if they are the ones I can use.

[KateWinslet]

@MyCentura

Imagine a dataset like this,

There are 3 columns,

Flower (Input feature) → Common name of the flower in a user text
Ground truth (Target) → Scientific name of the flower
Top5 Scientific names (Ground Truth top5) → top5 scientific names

I am just using a flower dataset but assuming this is a very niche dataset.

I would like to know how can we use rerank models. I don’t have much experience on this mycentura. I read a bit about cross encoders models doing this,
may someone let me know how to go about this ? I don’t want a LLM to do this. Bert like models if they are the ones I can use.

Absolutely, cross-encoder models like BERT can re-rank predictions. Train on your data to predict accurate scientific names from flower text inputs.