Graph-based pruning of redundant reflections in LLM chain-of-thought reduces average reasoning tokens by 42% while preserving or improving accuracy.
• Branch origin principle: new attempts must branch fromstill-valid ancestor products, not from negated/dead-end nodes
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Graph-Based Chain-of-Thought Pruning for Reducing Redundant Reflections in Reasoning LLMs
Graph-based pruning of redundant reflections in LLM chain-of-thought reduces average reasoning tokens by 42% while preserving or improving accuracy.