Instructions to use yifeihu/TFT-ID-1.0 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use yifeihu/TFT-ID-1.0 with Transformers:

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

pipe = pipeline("image-text-to-text", model="yifeihu/TFT-ID-1.0", trust_remote_code=True)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("yifeihu/TFT-ID-1.0", trust_remote_code=True)
model = AutoModelForMultimodalLM.from_pretrained("yifeihu/TFT-ID-1.0", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use yifeihu/TFT-ID-1.0 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "yifeihu/TFT-ID-1.0"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "yifeihu/TFT-ID-1.0",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/yifeihu/TFT-ID-1.0

SGLang

How to use yifeihu/TFT-ID-1.0 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 "yifeihu/TFT-ID-1.0" \
    --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": "yifeihu/TFT-ID-1.0",
		"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 "yifeihu/TFT-ID-1.0" \
        --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": "yifeihu/TFT-ID-1.0",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use yifeihu/TFT-ID-1.0 with Docker Model Runner:
```
docker model run hf.co/yifeihu/TFT-ID-1.0
```

What counts as "text"?

by mstachow - opened Aug 14, 2024

Discussion

mstachow

Aug 14, 2024

This model is awesome and is something I've been looking for for a very long time. I just have a question so I understand how to use it better - what did you consider as "text" when you were doing the original training data? For instance, did you put bounding boxes around logical components like paragraphs or sections? Was there any specific criteria you used?

yifeihu

Owner Aug 17, 2024

This comment has been hidden

yifeihu

Owner Aug 17, 2024

This model is awesome and is something I've been looking for for a very long time. I just have a question so I understand how to use it better - what did you consider as "text" when you were doing the original training data? For instance, did you put bounding boxes around logical components like paragraphs or sections? Was there any specific criteria you used?

Hi @mstachow ,
"text" are titles, section text, and other main sections that can be parsed with an OCR model later.
I excluded certain parts: page header, footer, footnotes, authors and affiliations.
The model will break a long text sections into multiple smaller ones. This is by design and the idea is to make the OCR text length more normalized.

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