A counterfactual code augmentation framework with concept-aware tuning is claimed to improve LLM understanding of code concepts, validated across multiple models and benchmarks.
IEEEtran_HOWTO
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Improving Code Understanding in Large Language Models through Concept-Aware Consistency Learning
A counterfactual code augmentation framework with concept-aware tuning is claimed to improve LLM understanding of code concepts, validated across multiple models and benchmarks.