PORTool improves accuracy and reduces tool calls in multi-tool LLM reasoning by using importance-aware policy optimization on rewarded rollout trees that compare alternative decisions sharing prefixes.
name": "knowledge_search
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
PORTool: Importance-Aware Policy Optimization with Rewarded Tree for Multi-Tool-Integrated Reasoning
PORTool improves accuracy and reduces tool calls in multi-tool LLM reasoning by using importance-aware policy optimization on rewarded rollout trees that compare alternative decisions sharing prefixes.