lytang/

LLM-AggreFact

We announce the LLM-AggreFact leaderboard with 35 latest fact-checking models being evaluated.

We include one additional dataset RAGTruth to our benchmark. We convert the dataset to the same format as in our benchmark and removed those non-checkworthy claims. We include a randomly sampled subset of the training set from RAGTruth into the validation set of the benchmark since the original training set is too large after conversion.… See the full description on the dataset page: https://huggingface.co/datasets/lytang/LLM-AggreFact.