Molmo 7B-D Model Card with Endpoint Usage

This is a copy of the original Molmo 7B-D model card with additional information about using the model via Hugging Face Inference Endpoints.

Using the Model via Inference Endpoints

Note: The following implementation is a community-contributed endpoint handler and is not an official implementation. For the official model and its usage, please refer to the official Molmo 7B-D model page.

You should see a Deploy via Inference Endpoints option at the top of this model card.

Currently, this handler uses bloat16 for inference. The original authors found some differences in results vs using float32 weights. I didn't find results that degraded much in my initial experiments, but I may change this implementation in the future.

If you've deployed the model using Hugging Face's Inference Endpoints with a community-contributed handler, you can use it with the following code:

import requests
import json
import base64
from IPython.display import Image, display

API_URL = YOUR_ENDPOINT_URL_HERE
headers = {
    "Accept" : "application/json",
    "Authorization": "Bearer hf_TOKEN_HERE",
    "Content-Type": "application/json" 
}

# Function to encode image to base64
def encode_image(image_path):
    with open(image_path, "rb") as image_file:
        return base64.b64encode(image_file.read()).decode('utf-8')

# Path to your local image file
image_path = "hf-logo-with-title.png"

# Display the image (if in a Jupyter notebook)
display(Image(filename=image_path))

# Encode the image
base64_image = encode_image(image_path)

# Prepare the payload
payload = {
    "inputs": {
        "image": base64_image,
        "text_prompt": "Describe this image in detail."
    }
}

# Make the POST request
response = requests.post(API_URL, headers=headers, json=payload)

# Check if the request was successful
if response.status_code == 200:
    # Parse the JSON response
    result = response.json()
    print(result)
else:
    print("Error:", response.status_code)
    print("Response:", response.text)

# Print some debug information
print("\nDebug Information:")
print(f"API URL: {API_URL}")
print(f"Image Path: {image_path}")
print(f"Payload size: {len(json.dumps(payload))} bytes")
print(f"Response status code: {response.status_code}")

Example output:

[{'generated_text': ' The image features a simple, cartoon-style emoji on the left side, set against a white background. The emoji is a yellow circle with a white outline, depicting a smiling face with black eyes and a red tongue sticking out. The face has two small yellow dots on its cheeks, giving it a cheerful expression. The emoji\'s hands are positioned in front of its chest, as if it is hugging itself. To the right of the emoji, in large, dark blue text, the words "Hugging Face" are displayed. The overall design is minimalistic, with the emoji and text being the only elements in the image.'}]

This code snippet demonstrates how to use the model with an image file, encode it to base64, and send it to the inference endpoint for processing. Make sure to replace "hf_TOKEN_HERE" with your actual Hugging Face API token.

Remember that this is a community implementation and may not reflect the most up-to-date or official way to use the model. For the latest official information and usage instructions, always refer to the official Molmo 7B-D model page.

Original Molmo 7B-D Model Card

The content below is a copy of the original model card. For the most up-to-date information, please refer to the official Molmo 7B-D model page.

Molmo 7B-D

Molmo is a family of open vision-language models developed by the Allen Institute for AI. Molmo models are trained on PixMo, a dataset of 1 million, highly-curated image-text pairs. It has state-of-the-art performance among multimodal models with a similar size while being fully open-source. You can find all models in the Molmo family here. Learn more about the Molmo family in our announcement blog post.

Molmo 7B-D is based on Qwen2-7B and uses OpenAI CLIP as vision backbone. It performs comfortably between GPT-4V and GPT-4o on both academic benchmarks and human evaluation. It powers the Molmo demo at molmo.allenai.org.

This checkpoint is a preview of the Molmo release. All artifacts used in creating Molmo (PixMo dataset, training code, evaluations, intermediate checkpoints) will be made available at a later date, furthering our commitment to open-source AI development and reproducibility.

Sign up here to be the first to know when artifacts are released.

Quick links:

💬 Demo
📂 All Models
📃 Paper
🎥 Blog with Videos

Quick Start

To run Molmo, first install dependencies:

pip install einops torchvision

Then, follow these steps:

from transformers import AutoModelForCausalLM, AutoProcessor, GenerationConfig
from PIL import Image
import requests

# load the processor
processor = AutoProcessor.from_pretrained(
    'allenai/Molmo-7B-D-0924',
    trust_remote_code=True,
    torch_dtype='auto',
    device_map='auto'
)

# load the model
model = AutoModelForCausalLM.from_pretrained(
    'allenai/Molmo-7B-D-0924',
    trust_remote_code=True,
    torch_dtype='auto',
    device_map='auto'
)

# process the image and text
inputs = processor.process(
    images=[Image.open(requests.get("https://picsum.photos/id/237/536/354", stream=True).raw)],
    text="Describe this image."
)

# move inputs to the correct device and make a batch of size 1
inputs = {k: v.to(model.device).unsqueeze(0) for k, v in inputs.items()}

# generate output; maximum 200 new tokens; stop generation when <|endoftext|> is generated
output = model.generate_from_batch(
    inputs,
    GenerationConfig(max_new_tokens=200, stop_strings="<|endoftext|>"),
    tokenizer=processor.tokenizer
)

# only get generated tokens; decode them to text
generated_tokens = output[0,inputs['input_ids'].size(1):]
generated_text = processor.tokenizer.decode(generated_tokens, skip_special_tokens=True)

# print the generated text
print(generated_text)

# >>>  This image features an adorable black Labrador puppy, captured from a top-down
#      perspective. The puppy is sitting on a wooden deck, which is composed ...

To make inference more efficient, run with autocast:

with torch.autocast(device_type="cuda", enabled=True, dtype=torch.bfloat16):
  output = model.generate_from_batch(
      inputs,
      GenerationConfig(max_new_tokens=200, stop_strings="<|endoftext|>"),
      tokenizer=processor.tokenizer
  )

We did most of our evaluation in this setting (autocast on, but float32 weights)

To even further reduce the memory requirements, the model can be run with bfloat16 weights:

model.to(dtype=torch.bfloat16)
inputs["images"] = inputs["images"].to(torch.bfloat16)
output = model.generate_from_batch(
    inputs,
    GenerationConfig(max_new_tokens=200, stop_strings="<|endoftext|>"),
    tokenizer=processor.tokenizer
)

Note that we have observed that this can change the output of the model compared to running with float32 weights.

Evaluations

Model	Average Score on 11 Academic Benchmarks	Human Preference Elo Rating
Molmo 72B	81.2	1077
Molmo 7B-D (this model)	77.3	1056
Molmo 7B-O	74.6	1051
MolmoE 1B	68.6	1032
GPT-4o	78.5	1079
GPT-4V	71.1	1041
Gemini 1.5 Pro	78.3	1074
Gemini 1.5 Flash	75.1	1054
Claude 3.5 Sonnet	76.7	1069
Claude 3 Opus	66.4	971
Claude 3 Haiku	65.3	999
Qwen VL2 72B	79.4	1037
Qwen VL2 7B	73.7	1025
Intern VL2 LLAMA 76B	77.1	1018
Intern VL2 8B	69.4	953
Pixtral 12B	69.5	1016
Phi3.5-Vision 4B	59.7	982
PaliGemma 3B	50.0	937
LLAVA OneVision 72B	76.6	1051
LLAVA OneVision 7B	72.0	1024
Cambrian-1 34B	66.8	953
Cambrian-1 8B	63.4	952
xGen - MM - Interleave 4B	59.5	979
LLAVA-1.5 13B	43.9	960
LLAVA-1.5 7B	40.7	951

Benchmarks: AI2D test, ChartQA test, VQA v2.0 test, DocQA test, InfographicVQA test, TextVQA val, RealWorldQA, MMMU val, MathVista testmini, CountBenchQA, Flickr Count (we collected this new dataset that is significantly harder than CountBenchQA).

FAQs

I'm getting an error a broadcast error when processing images!

Your image might not be in RGB format. You can convert it using the following code snippet:

from PIL import Image

image = Image.open(...)

if image.mode != "RGB":
    image = image.convert("RGB")

Molmo doesn't work great with transparent images!

We received reports that Molmo models might struggle with transparent images. For the time being, we recommend adding a white or dark background to your images before passing them to the model. The code snippet below shows how to do this using the Python Imaging Library (PIL):


# Load the image
url = "..."
image = Image.open(requests.get(url, stream=True).raw)

# Convert the image to grayscale to calculate brightness
gray_image = image.convert('L')  # Convert to grayscale

# Calculate the average brightness
stat = ImageStat.Stat(gray_image)
average_brightness = stat.mean[0]  # Get the average value

# Define background color based on brightness (threshold can be adjusted)
bg_color = (0, 0, 0) if average_brightness > 127 else (255, 255, 255)

# Create a new image with the same size as the original, filled with the background color
new_image = Image.new('RGB', image.size, bg_color)

# Paste the original image on top of the background (use image as a mask if needed)
new_image.paste(image, (0, 0), image if image.mode == 'RGBA' else None)

# Now you can pass the new_image to Molmo
processor = AutoProcessor.from_pretrained(
    'allenai/Molmo-7B-D-0924',
    trust_remote_code=True,
    torch_dtype='auto',
    device_map='auto'
)

License and Use

This model is licensed under Apache 2.0. It is intended for research and educational use. For more information, please see our Responsible Use Guidelines.

davanstrien
/

Molmo-7B-D-0924