tasks/text.py

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

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

#additional imports 
from transformers import Trainer, TrainingArguments, DistilBertForSequenceClassification, DistilBertTokenizerFast
import logging

router = APIRouter()

DESCRIPTION = "Random Baseline"
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"]

    tokenizer = DistilBertTokenizerFast.from_pretrained("distilbert-base-uncased")
    # Tokenize the datasets
    def tokenize_function(examples):
        return tokenizer(examples["quote"], padding="max_length", truncation=True)

    train_dataset = dataset["train"].map(tokenize_function, batched=True)
    test_dataset = dataset["test"].map(tokenize_function, batched=True)

    model = DistilBertForSequenceClassification.from_pretrained("distilbert-base-uncased", num_labels=8)  # Set num_labels for your classification task

    training_args = TrainingArguments(
    output_dir="./results",            
    eval_strategy="epoch",       # Evaluation strategy (can be "steps" or "epoch")
    per_device_train_batch_size=16,    # Batch size for training
    per_device_eval_batch_size=64,     # Batch size for evaluation
    num_train_epochs=3,                # Number of training epochs
    logging_dir="./logs",              # Directory for logs
    logging_steps=10,                  # How often to log
    )
    
    trainer = Trainer(
    model=model,                       # The model to train
    args=training_args,                # The training arguments
    train_dataset=train_dataset,       # The training dataset
    eval_dataset=test_dataset          # The evaluation dataset
    )
    
    
    trainer.train()
    predictions = trainer.evaluate()


    #--------------------------------------------------------------------------------------------
    # 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