Pre-Act adds explicit multi-step planning to ReAct agent prompts; a fine-tuned 70B Llama reached 0.92 action recall on Almita, beating GPT-4's 0.54.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Pre-Act: Multi-Step Planning and Reasoning Improves Acting in LLM Agents
Pre-Act adds explicit multi-step planning to ReAct agent prompts; a fine-tuned 70B Llama reached 0.92 action recall on Almita, beating GPT-4's 0.54.