aml-project-deploy / verify_models.py
Antigravity
Stable deployment version: Lazy loading and Docker optimized
2d802f0
import sys
import os
# Add current directory to path
sys.path.append(os.getcwd())
from models_loader import loader
import torch
from PIL import Image
import numpy as np
def test_models():
print("--- Starting Model Verification ---")
# 1. Sentiment
print("\nTesting Sentiment Analysis...")
if loader.sentiment_pipeline:
res = loader.sentiment_pipeline("I love this project!")
print(f"Result: {res}")
else:
print("FAILED: Sentiment pipeline not loaded")
# 2. QA
print("\nTesting Question Answering...")
if loader.qa_pipeline:
res = loader.qa_pipeline(question="What is this?", context="This is a test.")
print(f"Result: {res}")
else:
print("FAILED: QA pipeline not loaded")
# 3. Translation
print("\nTesting Translation (MT-EN-UR)...")
if loader.translator_pipeline:
res = loader.translator_pipeline("Hello, how are you?")
print(f"Result: {res}")
else:
print("FAILED: Translation pipeline not loaded")
# 4. Text Gen
print("\nTesting Text Generation...")
if loader.text_gen_pipeline:
res = loader.text_gen_pipeline("Once upon a time", max_length=20)
print(f"Result: {res}")
else:
print("FAILED: Text Gen pipeline not loaded")
# 5. ZSL
print("\nTesting Zero-Shot Learning...")
if loader.zsl_pipeline:
res = loader.zsl_pipeline("This is about sports.", candidate_labels=["politics", "sports", "cooking"])
print(f"Result: {res['labels'][0]}")
else:
print("FAILED: ZSL pipeline not loaded")
# 6. Gender Classifier (Mini)
print("\nTesting Image Classification (Gender)...")
if loader.gender_classifier:
# Create a dummy image
dummy_img = Image.fromarray(np.uint8(np.random.rand(224,224,3)*255))
res = loader.gender_classifier(dummy_img)
print(f"Result: {res}")
else:
print("FAILED: Gender classifier pipeline not loaded")
print("\n--- Verification Complete ---")
if __name__ == "__main__":
test_models()