model_config.json

{
    "pdf_processing": {
        "extract_images": false,
        "infer_table_structure": true,
        "strategy": "fast",
        "chunking_strategy": "by_title",
        "model_name": "yolox",
        "max_characters": 10000,
        "combine_text_under_n_chars": 100
      },
    "allowed_extensions": "pdf",
    "embeddings": "huggingface",
    "embeddings_model": "BAAI/bge-small-en-v1.5",
    "llm_model": "gpt-4",
    "model_temp": 0.2,
    "max_tokens": 512,
    "context_window": 5000,
    "UPLOAD_FOLDER": "../path/to/upload/folder",
    "GPT_PROMPT_PATH": "data/prompts/prompt_gpt.txt",
    "MISTRAL_PROMPT_PATH": "data/prompts/prompt_mistral.txt",
    "INFO_PROMPT_PATH": "data/prompts/prompt_info.txt",
    "peer_review_journals_path": "data/prompts/peer_review_journals.txt",
    "eq_network_journals_path": "data/prompts/eq_network_journals.txt",
    "queries": [
        "Is anything in the article shared (data, code)? Look for the words like Supporting, Additional, supplementary  information/code/material/datashar -ing/-ed/-e, available, reproducibility and similar  + links, appendix",
        "Has anything in the article been registered (in advance)?",
        "Does the article follow any reporting guidelines? To answer this question, follow the 3 steps sequentially. If any of the steps is true, assign a score of 1 and if all the steps are false, give a score of 0. STEP 1. Look for ISRCTN registry. STEP 2. Look if it is published in either The Lancet, The New England Journal of Medicine (NEJM), Journal of the American Medical Association (JAMA), British Medical Journal (BMJ), Annals of Internal Medicine, Nature Medicine, or Journal of Clinical Oncology. STEP 3. Look for one of the following guidelines, CONSORT for randomized controlled trials, PRISMA for meta‐analyses or systematic reviews, MOOSE for Meta-analyses of observational studies, STARD for diagnostic/prognostic studies, ARRIVE for animal pre-clinical studies, STROBE observational studies,SPIRIT for study protocols, CARE for case reports, AGREE for clinical practice guidelines, SRQR for qualitative researches,SQUIRE for quality improvement studies, SPIRIT Statement: Standard Protocol Items: Recommendations for Interventional Trials, PRIMER Collaboration: PRESENTATION AND INTERPRETATION OF MEDICAL RESEARCH, MIBBI: Minimum Information for Biological and Biomedical Investigations, COREQ: Consolidated Criteria for Reporting Qualitative Research, MDAR (Materials Design Analysis Reporting) reproducibility checklist is not a traditional reporting guideline like CONSORT or PRISMA. Instead, it's a tool designed to enhance the reproducibility and transparency of scientific research, REMARK (Reporting Recommendations for Tumor Marker Prognostic Studies).",
        "Is the methodology described in detail (where, when, how, what, who)?",
        "Is the data collection processes described in detail (where, when, how, what, who)?",
        "Is there any sample description? eg. size, demographics, recruitment, in-/exclusion criteria",
        "Does the article describe the data analysis process?",
        "Were measures taken to avoid or minimize systematic bias?",
        "Has the article been published in a journal?"],
    "criteria": [
        "Data and code sharing.",
        "Has anything in the article been registered (in advance)?",
        "Does the article follow any reporting guidelines?",
        "Description of methodology",
        "Data collection processes",
        "Sample description. eg. size, demographics, recruitment, in-/exclusion criteria",
        "Data analysis process",
        "Measures to minimize systematic bias",
        "Peer Review"],
    "journal_query": "Is the given research paper published in any of the following journals: {}?",
    "author_query": "Give me details about the institutions (like university or hospital) and contact details (eg. email) of the corresponding author."    
}