Large language models can parse compiler intermediate representations but consistently fail at instruction-level reasoning such as control flow, loop handling, and exact execution simulation.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Can Large Language Models Understand Intermediate Representations in Compilers?
Large language models can parse compiler intermediate representations but consistently fail at instruction-level reasoning such as control flow, loop handling, and exact execution simulation.