Pre-rollout task difficulty for agentic benchmarks is predictable from token-level entropy features, with Spearman rho=0.399 in-distribution and 0.225 out-of-distribution.
Mle-bench: Evaluating machine learning agents on machine learning engineering
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Predicting Task Difficulty Without Rollouts
Pre-rollout task difficulty for agentic benchmarks is predictable from token-level entropy features, with Spearman rho=0.399 in-distribution and 0.225 out-of-distribution.