Instructions to use vida-nyu/flan-t5-base-dataref-info-extract with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use vida-nyu/flan-t5-base-dataref-info-extract with Transformers:

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

pipe = pipeline("text-generation", model="vida-nyu/flan-t5-base-dataref-info-extract")

# Load model directly
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("vida-nyu/flan-t5-base-dataref-info-extract")
model = AutoModelForSeq2SeqLM.from_pretrained("vida-nyu/flan-t5-base-dataref-info-extract", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use vida-nyu/flan-t5-base-dataref-info-extract with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "vida-nyu/flan-t5-base-dataref-info-extract"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "vida-nyu/flan-t5-base-dataref-info-extract",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/vida-nyu/flan-t5-base-dataref-info-extract

SGLang

How to use vida-nyu/flan-t5-base-dataref-info-extract 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 "vida-nyu/flan-t5-base-dataref-info-extract" \
    --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": "vida-nyu/flan-t5-base-dataref-info-extract",
		"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 "vida-nyu/flan-t5-base-dataref-info-extract" \
        --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": "vida-nyu/flan-t5-base-dataref-info-extract",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use vida-nyu/flan-t5-base-dataref-info-extract with Docker Model Runner:
```
docker model run hf.co/vida-nyu/flan-t5-base-dataref-info-extract
```

Dataset Information Extraction from Scientific Text

Fine-tuned flan-t5-base to extract dataset identifiers and repository references from snippets of scientific publications.

What it does: Finds dataset IDs (GSE123456, DOIs, etc.) and repository names (GEO, Zenodo, etc.) in text snippets from papers.

Trained on: vida-nyu/pmc-articles-dataset-mentions-snippets - PMC article snippets extracted using Data Gatherer

Quick Start

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

model = AutoModelForSeq2SeqLM.from_pretrained("vida-nyu/flan-t5-base-dataref-info-extract")
tokenizer = AutoTokenizer.from_pretrained("vida-nyu/flan-t5-base-dataref-info-extract")

text = "Extract dataset information: The data are in GEO under GSE123456."
inputs = tokenizer(text, return_tensors="pt", max_length=512, truncation=True)
outputs = model.generate(**inputs, max_length=256, num_beams=4)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Training Details

Base: flan-t5-base (250M params)
Data: PMC snippets with/without dataset mentions. Used 80/20 train-test split
Config: 5 epochs, lr=3e-4, batch=16 (effective)
Input: Max 512 tokens with prefix "Extract dataset information: "
Output: Structured dataset info or null for no-dataset cases

Limitations

Max 512 tokens input (truncates long passages)
Trained on biomedical literature (may underperform on other domains)
English only
May miss novel/obscure repositories

Citation

@software{flan_t5_dataset_extraction,
  title={Flan-T5 for Dataset Information Extraction},
  author={VIDA Lab, NYU},
  year={2025},
  url={https://huggingface.co/vida-nyu/flan-t5-base-dataref-info-extract}
}

Contact: GitHub Issues | VIDA Lab, NYU

Downloads last month: 12

Safetensors

Model size

0.2B params

Tensor type

F32

Model tree for vida-nyu/flan-t5-base-dataref-info-extract

Base model

google/flan-t5-base

Finetuned

(922)

this model

vida-nyu
/

flan-t5-base-dataref-info-extract