"""Central configuration for the conspiracy language classifier."""

CONFIG = {
    # Model — XLM-RoBERTa-base: right-sized for ~20k samples, avoids overfitting
    "model_name": "xlm-roberta-base",
    "max_length": 256,
    "num_labels": 2,

    # LoRA
    "lora_r": 256,
    "lora_alpha": 512,
    "lora_dropout": 0.05,
    "lora_target_modules": ["query", "value"],

    # Training
    "learning_rate": 2e-5,
    "weight_decay": 0.01,
    "num_epochs": 4,
    "per_device_train_batch_size": 16,
    "gradient_accumulation_steps": 1,
    "warmup_ratio": 0.1,
    "max_grad_norm": 1.0,
    "fp16": "auto",
    "bf16": "auto",
    "gradient_checkpointing": False,

    # Early stopping
    "early_stopping_patience": 2,
    "early_stopping_metric": "eval_f1",

    # Data
    "covid_debias_prob": 0.4,
    "max_supplement_samples": 20000,
    "val_ratio": 0.1,
    "test_ratio": 0.1,
    "covid_test_size": 500,
    "seed": 42,

    # Paths
    "output_dir": "./results",
    "logging_dir": "./logs",
}

# COVID-specific terms grouped by category for debiasing
COVID_TERMS = {
    "disease_name": [
        "covid-19", "covid", "coronavirus", "sars-cov-2", "corona virus",
        "the virus", "the pandemic", "pandemic", "epidemic",
    ],
    "treatment": [
        "vaccine", "vaccination", "vaccines", "vaccinations", "vax", "vaxx",
        "pfizer", "moderna", "astrazeneca", "johnson & johnson", "j&j",
        "booster", "boosters", "mrna", "spike protein", "jab",
        "ivermectin", "hydroxychloroquine",
    ],
    "authority_figure": [
        "fauci", "dr. fauci", "dr fauci", "anthony fauci",
        "bill gates", "gates", "tedros",
    ],
    "organization": [
        "cdc", "who", "world health organization",
        "centers for disease control", "nih", "fda",
        "big pharma", "pharmaceutical companies",
    ],
    "location": [
        "wuhan", "china", "chinese", "wuhan lab",
    ],
    "measure": [
        "lockdown", "lockdowns", "quarantine", "mask mandate",
        "social distancing", "stay-at-home order", "shelter in place",
        "contact tracing",
    ],
    "concept": [
        "gain of function", "lab leak", "wet market", "china virus",
        "plandemic", "scamdemic", "herd immunity",
    ],
}

# Generic replacements per category
REPLACEMENTS = {
    "disease_name": [
        "the disease", "the illness", "the outbreak", "the health crisis",
        "the epidemic", "the infection",
    ],
    "treatment": [
        "the treatment", "the medicine", "the remedy", "the therapy",
        "the drug", "the cure", "the injection",
    ],
    "authority_figure": [
        "the official", "the director", "the leader", "the authority",
        "the public figure", "the expert",
    ],
    "organization": [
        "the agency", "the institution", "the organization",
        "the health authority", "the regulatory body",
    ],
    "location": [
        "the region", "the area", "the country", "the foreign lab",
    ],
    "measure": [
        "the policy", "the restriction", "the mandate", "the regulation",
        "the government order", "the safety measure",
    ],
    "concept": [
        "the cover-up", "the secret program", "the hidden agenda",
        "the conspiracy", "the scheme",
    ],
}