A fine-tuned Llama 3 model with a trained document critic and program-based reasoning beats GPT-4o and other baselines on a new carbon footprint QA benchmark.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
CF-RAG: A Dataset and Method for Carbon Footprint QA Using Retrieval-Augmented Generation
A fine-tuned Llama 3 model with a trained document critic and program-based reasoning beats GPT-4o and other baselines on a new carbon footprint QA benchmark.