[HANDS-ON BUG] Error in chat completion in unit 1 dummy agent library

[rahulsalgare]

Describe the bug
Getting error 'the requested model is not a chat model' while running the code given in dummy agent library

To Reproduce

import os
from huggingface_hub import InferenceClient

HF_TOKEN = os.environ.get("HF_TOKEN")

client = InferenceClient(model="meta-llama/Llama-4-Scout-17B-16E-Instruct")

output = client.chat.completions.create(
    messages=[
        {"role": "user", "content": "The capital of France is"},
    ],
    stream=False,
    max_tokens=20,
)
print(output.choices[0].message.content)

getting the error as,

Bad request:
{'message': "The requested model 'meta-llama/Llama-4-Scout-17B-16E-Instruct' is not a chat model.", 'type': 'invalid_request_error', 'param': 'model', 'code': 'model_not_supported'}

[Jason-liujc]

Was able to get around it by using another model

client = InferenceClient(model="meta-llama/Meta-Llama-3-8B-Instruct")

However, this doesn't let me reproduce the supposed response outlined here: https://huggingface.co/agents-course/notebooks/blob/main/unit1/dummy_agent_library.ipynb

[KongaraLikhith]

Why were we told to request access for the model? How's it linked to the `HF_TOKEN`?

In the dummy library code, the HF_TOKEN was declared but never used? are we going to use it in the future?

After getting access to the model and generating the token, do we need to do it like this?

client = InferenceClient(
    model="meta-llama/Llama-4-Scout-17B-16E-Instruct",
    token=HF_TOKEN
)

[Mahdyy02]

I used this model instead and it fixed the error for me:
meta-llama/Meta-Llama-3-70B-Instruct

The issue likely happened because meta-llama/Llama-4-Scout-17B-16E-Instruct is currently deployed on the Hugging Face router as a text-generation (completion) model rather than a chat-wrapped model.