Which model to use?

[avinregmi]

I've noticed that models are trained on STS data or NLI data. In the readme page it says that STS trained models are specifically well suited for semantic textual similarity. When should i use models trained on NLI dataset? Is there a specific application where NLI would be better?

[nreimers]

Hi @avinregmi I haven't tested it yet. But sometimes it is not that predicable which model works best for a specific use case