SciFact
A fact-checking benchmark dataset focused on scientific literature, requiring systems to verify the veracity of scientific claims against abstracts from scientific papers.
Timeline (last 90 days)
On the SciFact dataset (300 queries, 5,183 scientific abstracts, 768 dimensions), the int8+int4 repack model retained 97.9% of the original float32 version's nDCG@10 (0.836 vs 0.854).
The research team generated 470,400 responses on three datasets—FEVER, FEVEROUS, and SciFact—using two 8B-scale large language models (Qwen and Llama).
All Facts (2)
On the SciFact dataset (300 queries, 5,183 scientific abstracts, 768 dimensions), the int8+int4 repack model retained 97.9% of the original float32 version's nDCG@10 (0.836 vs 0.854).
50%UnverifiedThe research team generated 470,400 responses on three datasets—FEVER, FEVEROUS, and SciFact—using two 8B-scale large language models (Qwen and Llama).
50%