Data Science and AI Lab · Quiz 2 · Sep 2025

In text generation with GPT-2, what effect does increasing the temperature parameter generally have?

Fill in the missing function call to correctly compile and execute the LangGraph workflow. graph = StateGraph() # ... add nodes and edges ... app = graph.________() output = app.________({"query": "What is LangGraph?"})

In LangGraph, which of the following statements correctly describe the role of **State**? (Select all that apply)

What does the following line do? x = x.to('cuda')

You are training a neural network using PyTorch. The following code blocks are parts of a typical training loop. Arrange them in the correct order to form a valid training loop for one epoch. Make necessary assumptions and choose the correct option (A) outputs = model(inputs) loss = criterion(outputs, labels) (B) optimizer.step() (C) for inputs, labels in dataloader: (D) loss.backward() (E) optimizer.zero_grad()

When performing gradient accumulation in PyTorch, which of the following statements are true?

The parent node of a decision tree P has 50 samples belonging to class A and 50 samples belonging to class B from a total of 100 samples. After the split at node P, the left child L has the 50 points belonging to class B and the right child R has the 50 points belonging to class A. What is the absolute value of Entropy(P) - (Entropy(L) + Entropy(R)) ?

What will this code retrieve? vectorstore = FAISS.from_documents(chunks, embedding_model) results = vectorstore.similarity_search("product launch", k=3) print(type(results[0]))

A dataset df contains 20000 rows and 7 columns. The columns are named id, road_type, num_lanes, weather, speed_limit, area and public. The number of unique values taken by the columns are as follows: id - 20000, road_type - 4, num_lanes - 4, weather - 3, speed_limit - 25, area - 30, public - 2.The following transformations were done on the dataset. df = df.drop(columns=['id']) encoder = ColumnTransformer( transformers=[ ('road_type', OneHotEncoder(drop = None, handle_unknown='ignore'), ['road_type']), ('weather', OneHotEncoder(drop = None, handle_unknown='ignore'), ['weather']), ('area', OrdinalEncoder(handle_unknown="use_encoded_value",unknown_value=100), ['area']) ], remainder='passthrough' ) X_encoded = encoder.fit_transform(df) How many columns are present in the dataset X_encoded?

What is the output of the following snippet of code? import numpy as np A = np.array([ [1, 5, 9], [10, 2, 1], [0, 2, 3] ]) array_sum = A.sum(axis=1) print(array_sum)

x and y are 1D NumPy arrays, that is, x.ndim = 1 and y.ndim = 1. Also, x.shape == y.shape. What would happen if we run x @ y?

You are working with two datasets and trying to combine them to get two columns in the combined_dataset using the following code: from datasets import concatenate_datasets, load_dataset ds1 = load_dataset("imdb", split="train[:4000]") ds2 = load_dataset("csv", data_files="reviews.csv") combined_dataset = concatenate_datasets([ds1, ds2]) Assume the following: The ds1 dataset has columns: text (string), label (int). The ds2 dataset has columns: review (string), sentiment (int).

You need to add a new column char_count, representing the character count of the text column in your dataset. Which of the following code snippets will achieve this using the .map() method?

You notice your RAG system returns irrelevant results. Which change might help? # Current setup splitter = RecursiveCharacterTextSplitter(chunk_size=5000, chunk_overlap=500) retriever = vectorstore.as_retriever(search_kwargs={"k": 1})

What will be the output of the following code? from langchain_text_splitters import RecursiveCharacterTextSplitter text = "Hello world. How are you? I am fine." splitter = RecursiveCharacterTextSplitter(chunk_size=15, chunk_overlap=5) chunks = splitter.split_text(text) print(len(chunks))

In a standard 8-bit grayscale image, what pixel values typically represent pure black and pure white, respectively?

How is a standard color (RGB) image typically represented as a NumPy array?

Which of the following statements regarding Data Augmentation are correct?

What is the main difference between PIL.Image.open() and cv2.imread() when loading a color image?