Communication and Verification in LLM Agents towards Collaboration under Information Asymmetry (Arxiv)
Run Peng*, Ziqiao Ma*, Amy Pang, Sikai Li, Zhang Xi-Jia, Yingzhuo Yu, Cristian-Paul Bara, Joyce Chai
There are four *.jsonl files under train/ folder, which corresponds to training data for models with four different communicative action spaces. The chain-of-thought reasoning traces are generated by gpt4o given the current game state and… See the full description on the dataset page:
https://huggingface.co/datasets/Roihn/Einstein-Puzzles-Data.
