Huggingface

import weave
from huggingface_hub import InferenceClient

# Initialize Weave
weave.init(project_name="quickstart-huggingface")

# Initialize Hugging Face Inference Client
huggingface_client = InferenceClient(
    api_key=os.environ.get("HUGGINGFACE_TOKEN")
)

# Make a chat completion inference call to the Hugging Face Hub with the Llama-3.2-11B-Vision-Instruct model
image_url = "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
response = huggingface_client.chat_completion(
    model="meta-llama/Llama-3.2-11B-Vision-Instruct",
    messages=[
        {
            "role": "user",
            "content": [
                {"type": "image_url", "image_url": {"url": image_url}},
                {"type": "text", "text": "Describe this image in one sentence."},
            ],
        }
    ],
    max_tokens=500,
    seed=42,
)

import base64
from PIL import Image


def encode_image(pil_image):
    import io
    buffer = io.BytesIO()
    pil_image.save(buffer, format="JPEG")
    buffer.seek(0)
    encoded_image = base64.b64encode(buffer.read()).decode("utf-8")
    return f"data:image/jpeg;base64,{encoded_image}"


@weave.op
def generate_image(prompt: str):
    return huggingface_client.text_to_image(
        prompt=prompt,
        model="black-forest-labs/FLUX.1-schnell",
        num_inference_steps=4,
    )


@weave.op
def check_image_correctness(image: Image.Image, image_generation_prompt: str):
    return huggingface_client.chat_completion(
        model="meta-llama/Llama-3.2-11B-Vision-Instruct",
        messages=[
            {
                "role": "user",
                "content": [
                    {"type": "image_url", "image_url": {"url": encode_image(image)}},
                    {
                        "type": "text",
                        "text": f"Is this image correct for the prompt: {image_generation_prompt}? Answer with only one word: yes or no",
                    },
                ],
            }
        ],
        max_tokens=500,
        seed=42,
    ).choices[0].message.content


@weave.op
def generate_image_and_check_correctness(prompt: str):
    image = generate_image(prompt)
    return {
        "image": image,
        "is_correct": check_image_correctness(image, prompt),
    }


response = generate_image_and_check_correctness("A cute puppy")

import rich


class CityVisitRecommender(weave.Model):
    model: str
    temperature: float = 0.7
    max_tokens: int = 500
    seed: int = 42

    @weave.op()
    def predict(self, city: str) -> str:
        return huggingface_client.chat_completion(
            model=self.model,
            messages=[
                {
                    "role": "system",
                    "content": "You are a helpful assistant meant to suggest places to visit in a city",
                },
                {"role": "user", "content": city},
            ],
            max_tokens=self.max_tokens,
            temperature=self.temperature,
            seed=self.seed,
        ).choices[0].message.content


city_visit_recommender = CityVisitRecommender(
    model="meta-llama/Llama-3.2-11B-Vision-Instruct",
    temperature=0.7,
    max_tokens=500,
    seed=42,
)
rich.print(city_visit_recommender.predict("New York City"))
rich.print(city_visit_recommender.predict("Paris"))