validate.py

# validate_upload.py
from transformers import AutoModelForSequenceClassification, AutoTokenizer

# 1️⃣ Hugging Face repo name of your uploaded model
repo_name = "yashu100/Chatbot"  # replace with your repo

try:
    # 2️⃣ Load model and tokenizer directly from Hugging Face
    model = AutoModelForSequenceClassification.from_pretrained(repo_name)
    tokenizer = AutoTokenizer.from_pretrained(repo_name)
    
    print("✅ Model and tokenizer loaded successfully from Hugging Face!")

    # 3️⃣ Test with a sample input
    sample_text = "This is a test sentence."
    inputs = tokenizer(sample_text, return_tensors="pt")
    outputs = model(**inputs)
    
    print("Sample input processed successfully. Model output:")
    print(outputs.logits)

except Exception as e:
    print("❌ Failed to load the model. Error:")
    print(e)

# file: push_final_model.py

from transformers import AutoTokenizer, AutoModelForSequenceClassification
from huggingface_hub import HfApi, ModelCard, ModelCardData
import os

# --------------------------------------------------------------------------
# --- ACTION REQUIRED: CONFIGURE YOUR MODEL DETAILS HERE ---
# --------------------------------------------------------------------------
# 1. The local directory where your best final model is saved.
LOCAL_MODEL_DIR = "C:\\Users\\comte\\FINAL_90_PERCENT_MODEL" # Make sure this path is correct

# 2. The ID for your NEW, FINAL repository on the Hugging Face Hub.
#    Use a new name to guarantee a clean slate.
#    Format: "YourHuggingFaceUsername/YourNewModelName"
#
#    ---> CHANGE "YourUsername" AND "aura-sentiment-final" BELOW <---
HUB_MODEL_ID = "yashu100/aura-sentiment-final"

# 3. Your Hugging Face Username
#    This is needed to create the repository.
HF_USERNAME = "yashu100" # ---> CHANGE THIS <---

# 4. A commit message for the upload.
COMMIT_MESSAGE = "Final production upload with automated model card for Inference API"
# --------------------------------------------------------------------------


# --- Main Execution Script ---
if __name__ == "__main__":
    print("--- Starting Final, Automated Push to Hub Script ---")

    # --- Verify local model exists ---
    if not os.path.exists(LOCAL_MODEL_DIR):
        raise FileNotFoundError(f"Local model directory not found at '{LOCAL_MODEL_DIR}'.")
    
    # --- Load local model and tokenizer ---
    print(f"Loading model from: {LOCAL_MODEL_DIR}...")
    tokenizer = AutoTokenizer.from_pretrained(LOCAL_MODEL_DIR)
    model = AutoModelForSequenceClassification.from_pretrained(LOCAL_MODEL_DIR)
    print("Model and tokenizer loaded successfully.")

    # --- STEP 1: Programmatically create the model card with metadata ---
    print("\nCreating model card with required metadata...")
    card_data = ModelCardData(
        language="en",
        license="apache-2.0",
        # This is the single most important tag for the Inference API
        pipeline_tag="text-classification", 
        tags=["mental-health", "sentiment-analysis", "sih-project"],
        model_name=f"Aura Sentiment Analysis Model ({HUB_MODEL_ID.split('/')[1]})"
    )
    
    card = ModelCard.from_template(card_data)
    
    # --- STEP 2: Create the repository on the Hub ---
    api = HfApi()
    repo_url = api.create_repo(
        repo_id=HUB_MODEL_ID,
        repo_type="model",
        exist_ok=True # Don't error if it already exists, just update it
    )
    print(f"Repository '{HUB_MODEL_ID}' created or already exists on the Hub.")

    # --- STEP 3: Push the model, tokenizer, AND the new model card ---
    print(f"\nUploading all files to: {HUB_MODEL_ID}")
    
    # Save the generated model card to the local directory before upload
    card.save(os.path.join(LOCAL_MODEL_DIR, 'README.md'))
    
    # Upload the entire contents of the local directory
    api.upload_folder(
        folder_path=LOCAL_MODEL_DIR,
        repo_id=HUB_MODEL_ID,
        repo_type="model",
        commit_message=COMMIT_MESSAGE
    )

    print("\n" + "="*50)
    print("✅ SUCCESS! Your model is correctly deployed.")
    print(f"You can view your model at: {repo_url}")
    print("\nPlease wait 5-10 minutes for the Inference API to activate.")
    print("="*50)