tests/test_idefix2.py

import os
import torch
import requests

from io import BytesIO
from PIL import Image
from transformers import AutoModelForCausalLM, AutoProcessor

from transformers.image_utils import (
    to_numpy_array,
    PILImageResampling,
    ChannelDimension,
)
from transformers.image_transforms import resize, to_channel_dimension_format


API_TOKEN = os.getenv("HF_TOKEN")

DEVICE = torch.device("cuda")
PROCESSOR = AutoProcessor.from_pretrained(
    "HuggingFaceM4/tr_272_bis_opt_step_15000_merge",
    token=API_TOKEN,
)
MODEL = AutoModelForCausalLM.from_pretrained(
    "HuggingFaceM4/tr_272_bis_opt_step_15000_merge",
    token=API_TOKEN,
    trust_remote_code=True,
    torch_dtype=torch.bfloat16,
).to(DEVICE)
image_seq_len = MODEL.config.perceiver_config.resampler_n_latents
BOS_TOKEN = PROCESSOR.tokenizer.bos_token
BAD_WORDS_IDS = PROCESSOR.tokenizer(
    ["<image>", "<fake_token_around_image>"], add_special_tokens=False
).input_ids


def convert_to_rgb(image):
    # `image.convert("RGB")` would only work for .jpg images, as it creates a wrong background
    # for transparent images. The call to `alpha_composite` handles this case
    if image.mode == "RGB":
        return image

    image_rgba = image.convert("RGBA")
    background = Image.new("RGBA", image_rgba.size, (255, 255, 255))
    alpha_composite = Image.alpha_composite(background, image_rgba)
    alpha_composite = alpha_composite.convert("RGB")
    return alpha_composite


# The processor is the same as the Idefics processor except for the BILINEAR interpolation,
# so this is a hack in order to redefine ONLY the transform method
def custom_transform(x):
    x = convert_to_rgb(x)
    x = to_numpy_array(x)

    height, width = x.shape[:2]
    aspect_ratio = width / height
    if width >= height and width > 980:
        width = 980
        height = int(width / aspect_ratio)
    elif height > width and height > 980:
        height = 980
        width = int(height * aspect_ratio)
    width = max(width, 378)
    height = max(height, 378)

    x = resize(x, (height, width), resample=PILImageResampling.BILINEAR)
    x = PROCESSOR.image_processor.rescale(x, scale=1 / 255)
    x = PROCESSOR.image_processor.normalize(
        x,
        mean=PROCESSOR.image_processor.image_mean,
        std=PROCESSOR.image_processor.image_std,
    )
    x = to_channel_dimension_format(x, ChannelDimension.FIRST)
    x = torch.tensor(x)
    return x


def download_image(url):
    try:
        # Send a GET request to the URL to download the image
        response = requests.get(url)
        # Check if the request was successful (status code 200)
        if response.status_code == 200:
            # Open the image using PIL
            image = Image.open(BytesIO(response.content))
            # Return the PIL image object
            return image
        else:
            print(f"Failed to download image. Status code: {response.status_code}")
            return None
    except Exception as e:
        print(f"An error occurred: {e}")
        return None


# Create text token inputs
image_seq = "<image>" * image_seq_len

instruction = "What is this?"
# Create pixel inputs
image = download_image(
    "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
)


def ask_vlm(instruction, image):

    inputs = PROCESSOR.tokenizer(
        [
            f"{BOS_TOKEN}<fake_token_around_image>{image_seq}<fake_token_around_image>{instruction}",
        ],
        return_tensors="pt",
        add_special_tokens=False,
        padding=True,
    )

    raw_images = [
        [image],
    ]
    output_images = [
        [PROCESSOR.image_processor(img, transform=custom_transform) for img in img_list]
        for img_list in raw_images
    ]
    total_batch_size = len(output_images)
    max_num_images = max([len(img_l) for img_l in output_images])
    max_height = max([i.size(2) for img_l in output_images for i in img_l])
    max_width = max([i.size(3) for img_l in output_images for i in img_l])
    padded_image_tensor = torch.zeros(
        total_batch_size, max_num_images, 3, max_height, max_width
    )
    padded_pixel_attention_masks = torch.zeros(
        total_batch_size, max_num_images, max_height, max_width, dtype=torch.bool
    )
    for batch_idx, img_l in enumerate(output_images):
        for img_idx, img in enumerate(img_l):
            im_height, im_width = img.size()[2:]
            padded_image_tensor[batch_idx, img_idx, :, :im_height, :im_width] = img
            padded_pixel_attention_masks[batch_idx, img_idx, :im_height, :im_width] = (
                True
            )

    inputs["pixel_values"] = padded_image_tensor
    inputs["pixel_attention_mask"] = padded_pixel_attention_masks
    inputs = {k: v.to(DEVICE) for k, v in inputs.items()}

    generated_ids = MODEL.generate(
        **inputs, bad_words_ids=BAD_WORDS_IDS, max_new_tokens=10
    )
    generated_texts = PROCESSOR.batch_decode(generated_ids, skip_special_tokens=True)
    return generated_texts


print(ask_vlm(instruction, image))