contracrow.json

{
  "llm": "claude-3-5-sonnet-20240620",
  "llm_config": null,
  "summary_llm": "claude-3-5-sonnet-20240620",
  "summary_llm_config": null,
  "embedding": "hybrid-text-embedding-3-large",
  "embedding_config": null,
  "temperature": 0.0,
  "batch_size": 1,
  "texts_index_mmr_lambda": 1.0,
  "verbosity": 0,
  "answer": {
    "evidence_k": 30,
    "evidence_retrieval": true,
    "evidence_summary_length": "about 300 words",
    "evidence_skip_summary": false,
    "answer_max_sources": 15,
    "answer_length": "about 200 words, but can be longer",
    "max_concurrent_requests": 4,
    "answer_filter_extra_background": false
  },
  "parsing": {
    "chunk_size": 7000,
    "use_doc_details": true,
    "overlap": 250,
    "citation_prompt": "Provide the citation for the following text in MLA Format. Do not write an introductory sentence. If reporting date accessed, the current year is 2024\n\n{text}\n\nCitation:",
    "structured_citation_prompt": "Extract the title, authors, and doi as a JSON from this MLA citation. If any field can not be found, return it as null. Use title, authors, and doi as keys, author's value should be a list of authors. {citation}\n\nCitation JSON:",
    "disable_doc_valid_check": false,
    "chunking_algorithm": "simple_overlap"
  },
  "prompts": {
    "summary": "Summarize the excerpt below to help answer a question.\n\nExcerpt from {citation}\n\n----\n\n{text}\n\n----\n\nQuestion: {question}\n\nDo not directly answer the question, instead summarize to give evidence to help answer the question. Stay detailed; report specific numbers, equations, or direct quotes (marked with quotation marks). Reply \"Not applicable\" if the excerpt is irrelevant. At the end of your response, provide an integer score from 1-10 on a newline indicating relevance to question. Do not explain your score.\n\nRelevant Information Summary ({summary_length}):",
    "qa": "Determine if the claim below is contradicted by the context below\n\n\n{context}\n\n----\n\nClaim: {question}\n\n\nDetermine if the claim is contradicted by the context. For each part of your response, indicate which sources most support it via citation keys at the end of sentences, like (Example2012Example pages 3-4). Only cite from the context below and only use the valid keys.\n\nRespond with the following XML format:\n\n<response>\n  <reasoning>...</reasoning>\n  <label>...</label>\n</response>\n\n\nwhere `reasoning` is your reasoning ({answer_length}) about if the claim is being contradicted. `label` is one of the following (must match exactly): \n\nexplicit contradiction\nstrong contradiction\ncontradiction\nnuanced contradiction\npossibly a contradiction\nlack of evidence\npossibly an agreement\nnuanced agreement\nagreement\nstrong agreement\nexplicit agreement\n\nDon't worry about other contradictions or agreements in the context, only focus on the specific claim. If there is no evidence for the claim, you should choose lack of evidence.",
    "select": "Select papers that may help answer the question below. Papers are listed as $KEY: $PAPER_INFO. Return a list of keys, separated by commas. Return \"None\", if no papers are applicable. Choose papers that are relevant, from reputable sources, and timely (if the question requires timely information).\n\nQuestion: {question}\n\nPapers: {papers}\n\nSelected keys:",
    "pre": null,
    "post": null,
    "system": "Answer in a direct and concise tone. Your audience is an expert, so be highly specific. If there are ambiguous terms or acronyms, first define them.",
    "use_json": true,
    "summary_json": "Excerpt from {citation}\n\n----\n\n{text}\n\n----\n\nQuestion: {question}\n\n",
    "summary_json_system": "Provide a summary of the relevant information that could help determine if a claim is contradicted or supported by this excerpt. The excerpt may be irrelevant. Do not directly answer if it is contradicted - only summarize relevant information. Respond with the following JSON format:\n\n{{\n  \"summary\": \"...\",\n  \"relevance_score\": \"...\"\n}}\n\nwhere `summary` is relevant information from excerpt ({summary_length}) and `relevance_score` is the relevance of `summary` to support or contradict the claim (integer out of 10). If any string entry in the JSON has newlines, be sure to escape them. "
  },
  "agent": {
    "agent_llm": "gpt-4o-2024-08-06",
    "agent_type": "ToolSelector",
    "agent_system_prompt": "You are a helpful AI assistant.",
    "agent_prompt": "Answer question: {question}\n\nSearch for papers, gather evidence, collect papers cited in evidence then re-gather evidence, answer, and complete. Gathering evidence will do nothing if you have not done a new search or collected new papers. If you do not have enough evidence to generate a good answer, you can:\n- Search for more papers (preferred)\n- Collect papers cited by previous evidence (preferred)\n- Gather more evidence using a different phrase\nIf you search for more papers or collect new papers cited by previous evidence, remember to gather evidence again. Once you have five or more pieces of evidence from multiple sources, or you have tried a few times, call the {complete_tool_name} tool to terminate. The current status of evidence/papers/cost is {status}",
    "search_count": 12,
    "timeout": 500.0,
    "tool_names": null
  }
}