A multi-agent pipeline generates benchmarks with rich metadata and reliable ground truths from textbooks, yielding three new sets in ML and finance domains that show lower expert-verified error rates and more uniform coverage than MMLU or GSM8K.
solution_graph
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Fine-Grained Benchmark Generation for Comprehensive Evaluation of Foundation Models
A multi-agent pipeline generates benchmarks with rich metadata and reliable ground truths from textbooks, yielding three new sets in ML and finance domains that show lower expert-verified error rates and more uniform coverage than MMLU or GSM8K.