Model generating claim in LLM aggrefact

[sanjanaramprasad]

Hi liyan

Thanks so much for your work and for releasing the resources! Is it possible to update the aggregate LLM dataset so that each row has the model identifier responsible for generating a claim? I plan to use this dataset for a few experiments on a metric and would appreciate being able to do a model/size analysis.

[Liyan06]

Hi,

I can try to add this at later time (probably in 1 or 2 weeks). This can be done by using sentence/word overlap to determine which model and corresponding benchmark dataset the claim originates from. If you have those mappings ealier than that, please let me know and I'll update the benchmark accordingly. Thanks!

[sanjanaramprasad]

Hi Liyan,

Just checking in if you were able to update the benchmark. I think it may be cleaner for you to do it since you processed the datasets. I had some issues tracing back to the models.