A predictor-based agent generates LLM pruning policies in seconds and reports large perplexity reductions on Llama2-7B and Llama3-8B.
J.; Rodgers, P.; Kilpatrick, P.; Spence, I
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
-
Beyond Manually Designed Pruning Policies with Second-Level Performance Prediction: A Pruning Framework for LLMs
A predictor-based agent generates LLM pruning policies in seconds and reports large perplexity reductions on Llama2-7B and Llama3-8B.