Connecting to open-source models in LangChain - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ component_behavior

intermediate

What is the output of this LangChain code connecting to an open-source model?

Consider this Python code using LangChain to connect to an open-source model with HuggingFaceHub. What will be printed?

LangChain

from langchain import HuggingFaceHub

repo_id = "google/flan-t5-small"
hf = HuggingFaceHub(repo_id=repo_id, model_kwargs={"temperature":0})

response = hf.predict("Translate English to French: 'Hello, how are you?'" )
print(response)

A "Bonjour, comment ça va ?"

B "Hello, how are you?"

C SyntaxError due to missing import

D TypeError because predict method does not exist

💡 Hint

Think about what the model is designed to do and what the predict method returns.

The HuggingFaceHub client calls the model to translate the input text. The output is the French translation of the English sentence.

📝 Syntax

intermediate

Which option correctly initializes a LangChain HuggingFaceHub client for an open-source model?

Choose the correct way to create a HuggingFaceHub instance for the model "facebook/bart-large-cnn" with temperature 0.7.

A HuggingFaceHub(model_id="facebook/bart-large-cnn", temperature=0.7)

B HuggingFaceHub(repo_id="facebook/bart-large-cnn", model_kwargs={"temperature":0.7})

C HuggingFaceHub(repo="facebook/bart-large-cnn", model_kwargs={temperature:0.7})

D HuggingFaceHub(repo_id="facebook/bart-large-cnn", temperature=0.7)

💡 Hint

Check the parameter names and how model_kwargs is passed as a dictionary.

The correct parameter for the model identifier is repo_id. The temperature must be inside model_kwargs as a dictionary.

🔧 Debug

advanced

What error does this LangChain code raise when connecting to an open-source model?

Analyze this code snippet and identify the error it will raise.

LangChain

from langchain import HuggingFaceHub

hf = HuggingFaceHub(repo_id="google/flan-t5-small")
response = hf.predict()
print(response)

A ValueError: repo_id not found

B AttributeError: 'HuggingFaceHub' object has no attribute 'predict'

C RuntimeError: Model loading failed

D TypeError: predict() missing 1 required positional argument: 'input'

💡 Hint

Check the method signature of predict and what arguments it requires.

The predict method requires an input string argument. Calling it without arguments raises a TypeError.

❓ state_output

advanced

What is the value of the variable 'output' after running this LangChain code?

Given this code connecting to an open-source model, what will output contain?

LangChain

from langchain import HuggingFaceHub

hf = HuggingFaceHub(repo_id="google/flan-t5-small", model_kwargs={"temperature":0})
output = hf.predict("Summarize: LangChain is a framework for building applications with LLMs.")

A "LangChain is a framework to build applications using large language models."

B "LangChain is a tool for data visualization."

C "Error: Model not found"

D "LangChain is a programming language."

💡 Hint

Think about what the model is expected to do with the input prompt.

The model summarizes the input text. The output is a concise summary related to LangChain and LLMs.

🧠 Conceptual

expert

Which option best explains why LangChain uses 'model_kwargs' when connecting to open-source models?

Why does LangChain require passing parameters inside model_kwargs when creating a HuggingFaceHub client?

A Because LangChain only supports models that require temperature tuning

B Because the model cannot run without <code>model_kwargs</code> being non-empty

C To allow flexible configuration of model behavior without changing the client interface

D To enforce strict typing on all model parameters at runtime

💡 Hint

Consider how passing a dictionary of parameters helps with different models.

Using model_kwargs lets LangChain pass any number of optional parameters to the model flexibly, adapting to different model APIs.