Instructions to use microsoft/udop-large-512-300k with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use microsoft/udop-large-512-300k with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="microsoft/udop-large-512-300k")

# Load model directly
from transformers import AutoProcessor, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("microsoft/udop-large-512-300k")
model = AutoModelForImageTextToText.from_pretrained("microsoft/udop-large-512-300k")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use microsoft/udop-large-512-300k with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "microsoft/udop-large-512-300k"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "microsoft/udop-large-512-300k",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/microsoft/udop-large-512-300k

SGLang

How to use microsoft/udop-large-512-300k with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "microsoft/udop-large-512-300k" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "microsoft/udop-large-512-300k",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "microsoft/udop-large-512-300k" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "microsoft/udop-large-512-300k",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use microsoft/udop-large-512-300k with Docker Model Runner:
```
docker model run hf.co/microsoft/udop-large-512-300k
```

How to fine tune on DocVQA

by hal9003 - opened Mar 6, 2024

Discussion

hal9003

Mar 6, 2024

Is there a tutorial or tips for fine tuning on QA task?

Mit1208

Mar 14, 2024

https://github.com/NielsRogge/Transformers-Tutorials/blob/master/UDOP/Fine_tune_UDOP_on_a_custom_dataset_(toy_RVL_CDIP_dataset).ipynb

I think you just need to update "target_sequence" with your question and answer. You can try with that.

nielsr

Mar 17, 2024

Indeed, and you could set the labels of the question tokens to -100 to make sure the model only learns to complete the question. Refer to my Donut example notebook.

hal9003

Mar 18, 2024

Thanks!

jainamhdoshi

Mar 19, 2024

ImportError: cannot import name 'UdopForConditionalGeneration' from 'transformers'
I am getting this error while importing it from transformers

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment