REST API inference in PyTorch - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load the PyTorch model for inference.

PyTorch

import torch

model = torch.load('[1]')
model.eval()

Drag options to blanks, or click blank then click option'

A'config.json'

B'train.py'

C'model.pth'

D'data.csv'

Attempts:

3 left

💡 Hint

Common Mistakes

Trying to load a script file instead of the model file.

Using a data file instead of the model file.

The model file is usually saved with a .pth extension. Loading 'model.pth' loads the trained model.

2fill in blank

medium

Complete the code to convert input JSON data into a PyTorch tensor for inference.

PyTorch

import torch
import json

input_json = '{"data": [1.0, 2.0, 3.0]}'
input_data = json.loads(input_json)
input_tensor = torch.tensor([1])

Drag options to blanks, or click blank then click option'

Ainput_data['data']

Binput_json['data']

Cinput_data['input']

Dinput_json['input']

Attempts:

3 left

💡 Hint

Common Mistakes

Trying to index the JSON string instead of the parsed dictionary.

Using the wrong key name.

After loading JSON, the data is accessed by the key 'data' in the dictionary input_data.

3fill in blank

hard

Fix the error in the code to perform inference and get the output tensor.

PyTorch

with torch.no_grad():
    output = model([1])
    prediction = output.argmax(dim=1).item()

Drag options to blanks, or click blank then click option'

Ainput_data

Binput_tensor

Cinput_json

Dinput_tensor.numpy()

Attempts:

3 left

💡 Hint

Common Mistakes

Passing raw JSON or dictionary instead of tensor.

Converting tensor to numpy array before passing to model.

The model expects a tensor input, so we pass input_tensor to the model.

4fill in blank

hard

Fill both blanks to create a Flask route that accepts POST requests and returns JSON predictions.

PyTorch

from flask import Flask, request, jsonify

app = Flask(__name__)

@app.route('/predict', methods=[[1]])
def predict():
    data = request.get_json()
    input_tensor = torch.tensor(data['input'])
    with torch.no_grad():
        output = model(input_tensor)
        pred = output.argmax(dim=1).item()
    return jsonify([2])

Drag options to blanks, or click blank then click option'

A'POST'

B'GET'

C{'prediction': pred}

D{'result': pred}

Attempts:

3 left

💡 Hint

Common Mistakes

Using GET method which does not send JSON body.

Returning JSON with unclear or wrong keys.

The route should accept POST requests, and the JSON response should have key 'prediction' with the predicted value.

5fill in blank

hard

Fill all three blanks to complete the Flask app that runs on port 5000 with debug mode on.

PyTorch

if __name__ == '__main__':
    app.run(host=[1], port=[2], debug=[3])

Drag options to blanks, or click blank then click option'

A'0.0.0.0'

B5000

CTrue

D'localhost'

Attempts:

3 left

💡 Hint

Common Mistakes

Using '0.0.0.0' which exposes the server externally.

Setting debug to False or a string instead of boolean True.

The app runs on localhost at port 5000 with debug mode enabled (True).