tasks/text.py

import random
from datetime import datetime

from datasets import load_dataset
from fastapi import APIRouter
from sklearn.metrics import accuracy_score

from .utils.emissions import clean_emissions_data, get_space_info, tracker
from .utils.evaluation import TextEvaluationRequest

router = APIRouter()

DESCRIPTION = "RF overfitted like crazy"
ROUTE = "/text"


@router.post(ROUTE, tags=["Text Task"], description=DESCRIPTION)
async def evaluate_text(request: TextEvaluationRequest):
    """
    Evaluate text classification for climate disinformation detection.

    Current Model: Random Baseline
    - Makes random predictions from the label space (0-7)
    - Used as a baseline for comparison
    """
    # Get space info
    username, space_url = get_space_info()

    # Define the label mapping
    LABEL_MAPPING = {
        "0_not_relevant": 0,
        "1_not_happening": 1,
        "2_not_human": 2,
        "3_not_bad": 3,
        "4_solutions_harmful_unnecessary": 4,
        "5_science_unreliable": 5,
        "6_proponents_biased": 6,
        "7_fossil_fuels_needed": 7,
    }

    # Load and prepare the dataset
    dataset = load_dataset(request.dataset_name)

    # Convert string labels to integers
    dataset = dataset.map(lambda x: {"label": LABEL_MAPPING[x["label"]]})

    # Split dataset
    train_test = dataset["train"].train_test_split(test_size=request.test_size, seed=request.test_seed)
    test_dataset = train_test["test"]

    # Start tracking emissions
    tracker.start()
    tracker.start_task("inference")

    # --------------------------------------------------------------------------------------------
    # YOUR MODEL INFERENCE CODE HERE
    # Update the code below to replace the random baseline by your model inference within the inference pass where the energy consumption and emissions are tracked.
    # --------------------------------------------------------------------------------------------

    # Make random predictions (placeholder for actual model inference)
    true_labels = test_dataset["label"]
    import joblib
    from sentence_transformers import SentenceTransformer

    embedding_model = SentenceTransformer("all-MiniLM-L6-v2")

    # Convert "quote" key into embeddings
    def embed_quote(example):
        example["quote_embedding"] = embedding_model.encode(example["quote"]).tolist()
        return example

    test_dataset = test_dataset.map(embed_quote, batched=True)
    # Load the model from the file
    rf_loaded = joblib.load("models/random_forest_model.pkl")
    import numpy as np

    # predictions = np.array([np.argmax(x) for x in preds[0]])
    predictions = rf_loaded.predict(np.array(test_dataset["quote_embedding"]))

    # --------------------------------------------------------------------------------------------
    # YOUR MODEL INFERENCE STOPS HERE
    # --------------------------------------------------------------------------------------------

    # Stop tracking emissions
    emissions_data = tracker.stop_task()

    # Calculate accuracy
    accuracy = accuracy_score(true_labels, predictions)

    # Prepare results dictionary
    results = {
        "username": username,
        "space_url": space_url,
        "submission_timestamp": datetime.now().isoformat(),
        "model_description": DESCRIPTION,
        "accuracy": float(accuracy),
        "energy_consumed_wh": emissions_data.energy_consumed * 1000,
        "emissions_gco2eq": emissions_data.emissions * 1000,
        "emissions_data": clean_emissions_data(emissions_data),
        "api_route": ROUTE,
        "dataset_config": {
            "dataset_name": request.dataset_name,
            "test_size": request.test_size,
            "test_seed": request.test_seed,
        },
    }

    return results