Spaces:

autoevaluate

/

model-evaluator

like 174

Runtime error

The current Zero Shot Text classification model is a fantastic way to evaluate the capabilities of LLMs on new datasets! With models like FLAN-T5 now hosted on the Hub, would it be possible to have a variant of the Zero Shot task that enables Chain of Thought prompting?

It would be awesome to be able to compare the scaling behavior of CoT reasoning with Instruction Finetuned models vs. Pure Autoregressive Models such as Bloom!