maheshcr/saslm-eval download history
maheshcr/saslm-eval is a multiple choice dataset on the Hugging Face Hub. In the last 30 days it was downloaded 62 times (6 in the last 7 days), and 62 times in total. It ranks #159,286 among datasets by monthly downloads.
SASLM-Eval A 466-item benchmark testing whether a language model has learned the ontology of Sri Aurobindo's philosophical writing, as opposed to its surface style. Perplexity measures fluency. This benchmark was built because perplexity on this corpus turned out not to be informative