Add configs for contracrow + wikicrow

paperqa/configs/contracrow.json

@@ -0,0 +1,17 @@
+{
+  "llm": "claude-3-5-sonnet-20240620",
+  "summary_llm": "claude-3-5-sonnet-20240620",
+  "agent": { "agent_type": "OpenAIFunctionsAgent", "search_count": 12 },
+  "answer": {
+    "evidence_summary_length": "about 300 words",
+    "answer_max_sources": 15,
+    "evidence_k": 30
+  },
+  "prompts": {
+    "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. ",
+    "use_json": true,
+    "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."
+  },
+  "embedding": "hybrid-text-embedding-3-large",
+  "parsing": { "chunk_size": 7000, "overlap": 250 }
+}

paperqa/configs/wikicrow.json

@@ -0,0 +1,22 @@
+{
+  "llm": "gpt-4-turbo-2024-04-09",
+  "summary_llm": "gpt-4-turbo-2024-04-09",
+  "agent": {
+    "agent_llm": "gpt-4-turbo-2024-04-09",
+    "agent_type": "OpenAIFunctionsAgent",
+    "search_count": 12
+  },
+  "answer": {
+    "evidence_summary_length": "about 300 words",
+    "answer_max_sources": 12,
+    "evidence_k": 25
+  },
+  "prompts": {
+    "summary_json_system": "Provide a summary of the relevant information that could help answer the question based on the excerpt. The excerpt may be irrelevant. Do not directly answer the question - only summarize relevant information. \n\nRespond with the following JSON format:\n\n{{\n  \"summary\": \"...\",\n  \"relevance_score\": \"...\",\n  \"gene_name: \"...\"\n}}\n\nwhere `summary` is relevant information from text - {summary_length}, \n`gene_name` is the gene discussed in the excerpt (may be different than query), and `relevance_score` is the relevance of `summary` to answer the question (integer out of 10)",
+    "use_json": true,
+    "qa": "Answer the question below with the context.\n\nContext (with relevance scores):\n\n{context}\n\n----\n\nQuestion: {question}\n\nWrite an answer based on the context. If the context provides insufficient information and the question cannot be directly answered, reply \"I cannot answer.\" For each part of your answer, 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. Write in the style of a Wikipedia article, with concise sentences and coherent paragraphs. The context comes from a variety of sources and is only a summary, so there may inaccuracies or ambiguities. Make sure the gene_names exactly match the gene name in the question before using a context. If quotes are present and relevant, use them in the answer. This answer will go directly onto Wikipedia, so do not add any extraneous information. Do not reference this prompt or your context. Do not include general summary information, it will be provided in an \"Overview\" section. Do not include the section title in your output. Avoid using adverb phrases like \"furthermore\", \"additionally\", and \"moreover\".\n\nAnswer ({answer_length}):",
+    "system": "Answer in a direct and concise tone."
+  },
+  "embedding": "hybrid-text-embedding-3-small",
+  "parsing": { "chunk_size": 7000, "overlap": 1750 }
+}