| {"cell_id": "q115_hard_negative_k4", "doc_id": "2023.findings-emnlp.248.pdf", "question": "In the paper main results table comparing Standard, Chain-of-Thought, and the proposed method across the five logical reasoning datasets, on which dataset does the proposed method (without self-refinement) achieve its highest accuracy when GPT-4 is the underlying LLM?", "answer": "LogicalDeduction", "answer_format": "Str", "condition": "hard_negative", "k": 4, "source_position": "first", "pdf_paths": ["/Users/jasonluo08/Desktop/AI Brain/algoverse_research/pdfs/2023.findings-emnlp.248.pdf", "/Users/jasonluo08/Desktop/AI Brain/algoverse_research/hard_negative_pdfs/web_39b6cadc4e9ca03e.pdf", "/Users/jasonluo08/Desktop/AI Brain/algoverse_research/hard_negative_pdfs/web_949ca008fd6aab55.pdf", "/Users/jasonluo08/Desktop/AI Brain/algoverse_research/hard_negative_pdfs/web_56044d2e22a63da0.pdf", "/Users/jasonluo08/Desktop/AI Brain/algoverse_research/hard_negative_pdfs/web_992a098addf13dd9.pdf"], "pdf_names": ["2023.findings-emnlp.248.pdf", "web_39b6cadc4e9ca03e.pdf", "web_949ca008fd6aab55.pdf", "web_56044d2e22a63da0.pdf", "web_992a098addf13dd9.pdf"], "n_pages": 107, "est_tokens": 96300, "expected_route": "openrouter", "bundle_filenames": ["source.pdf", "hn_1.pdf", "hn_2.pdf", "hn_3.pdf", "hn_4.pdf"], "original_filenames": ["2023.findings-emnlp.248.pdf", "web_39b6cadc4e9ca03e.pdf", "web_949ca008fd6aab55.pdf", "web_56044d2e22a63da0.pdf", "web_992a098addf13dd9.pdf"]} |
