SAGE reports 97-99% exact-match accuracy on GSM8K level-1 by triggering LoRA fine-tuning on flagged samples, but the flagging uses ground-truth answers and the final evaluation is not held out.
J.; Moulavi, D.; and Sander, J
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
A Lightweight Framework for Trigger-Guided LoRA-Based Self-Adaptation in LLMs
SAGE reports 97-99% exact-match accuracy on GSM8K level-1 by triggering LoRA fine-tuning on flagged samples, but the flagging uses ground-truth answers and the final evaluation is not held out.