PhageBench is the first benchmark for LLMs on raw bacteriophage genomes, with 5,600 samples across five tasks showing models beat random baselines on basic identification but fail on complex long-range reasoning.
Cheng Peng, Jiayu Shang, Jiaojiao Guan, Donglin Wang, and Yanni Sun
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
PhageBench: Can LLMs Understand Raw Bacteriophage Genomes?
PhageBench is the first benchmark for LLMs on raw bacteriophage genomes, with 5,600 samples across five tasks showing models beat random baselines on basic identification but fail on complex long-range reasoning.