Instructions to use abertsch/unlimiformer-bart-summscreen-earlyk with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use abertsch/unlimiformer-bart-summscreen-earlyk with Transformers:

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

pipe = pipeline("text-generation", model="abertsch/unlimiformer-bart-summscreen-earlyk")

# Load model directly
from transformers import AutoTokenizer, AutoModel

tokenizer = AutoTokenizer.from_pretrained("abertsch/unlimiformer-bart-summscreen-earlyk")
model = AutoModel.from_pretrained("abertsch/unlimiformer-bart-summscreen-earlyk")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use abertsch/unlimiformer-bart-summscreen-earlyk with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/abertsch/unlimiformer-bart-summscreen-earlyk

SGLang

How to use abertsch/unlimiformer-bart-summscreen-earlyk 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 "abertsch/unlimiformer-bart-summscreen-earlyk" \
    --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": "abertsch/unlimiformer-bart-summscreen-earlyk",
		"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 "abertsch/unlimiformer-bart-summscreen-earlyk" \
        --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": "abertsch/unlimiformer-bart-summscreen-earlyk",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use abertsch/unlimiformer-bart-summscreen-earlyk with Docker Model Runner:
```
docker model run hf.co/abertsch/unlimiformer-bart-summscreen-earlyk
```

YAML Metadata Warning:The pipeline tag "text2text-generation" is not in the official list: text-classification, token-classification, table-question-answering, question-answering, zero-shot-classification, translation, summarization, feature-extraction, text-generation, fill-mask, sentence-similarity, text-to-speech, text-to-audio, automatic-speech-recognition, audio-to-audio, audio-classification, audio-text-to-text, voice-activity-detection, depth-estimation, image-classification, object-detection, image-segmentation, text-to-image, image-to-text, image-to-image, image-to-video, unconditional-image-generation, video-classification, reinforcement-learning, robotics, tabular-classification, tabular-regression, tabular-to-text, table-to-text, multiple-choice, text-ranking, text-retrieval, time-series-forecasting, text-to-video, image-text-to-text, image-text-to-image, image-text-to-video, visual-question-answering, document-question-answering, zero-shot-image-classification, graph-ml, mask-generation, zero-shot-object-detection, text-to-3d, image-to-3d, image-feature-extraction, video-text-to-text, keypoint-detection, visual-document-retrieval, any-to-any, video-to-video, other

Model from the preprint Unlimiformer: Long-Range Transformers with Unlimited Length Input.

This model was finetuned from a BART-base model using Unlimiformer-aware early stopping, described in section 3.1 of the paper. It was finetuned on the dataset SummScreen using the data preprocessing pipeline from SLED; to load the validation or test set for use with these model, please use the datasets urialon/summ_screen_validation and urialon/summ_screen_test.

This is generally a weaker model than the retrieval-trained model and a stronger model than the baseline.

The inference demo is disabled because you must add the Unlimiformer files to your repo before this model can handle unlimited length input! See the Unlimiformer GitHub for setup instructions.

Downloads last month: 10

Datasets used to train abertsch/unlimiformer-bart-summscreen-earlyk

Paper for abertsch/unlimiformer-bart-summscreen-earlyk

Unlimiformer: Long-Range Transformers with Unlimited Length Input

Paper • 2305.01625 • Published May 2, 2023 • 6