Auto training and fast deployment for state-of-the-art NLP models

Announcing AutoNLP: an automatic way to train, evaluate and deploy state-of-the-art NLP models for different tasks.

Join beta

with your Hugging Face account

  • huggingface@autonlp:~
    # Upload your model data
    autonlp upload --project sentiment_detection --split train \
                   --col_mapping review:text,sentiment:target \
                   --files ~/datasets/train.csv
    # Train your model
    autonlp train --project sentiment_detection
    # Use your model 
    curl -X POST \
    		-H "Authorization: Bearer api_jeZrkpoqfjzioaRaerjlbRQeKykrop" \
    		-H "Content-Type: application/json" \
    		-d '{"inputs":"The goal of life is [MASK]."}' \
  • What is AutoNLP?
    AutoNLP is an automatic way to train and deploy state-of-the-art NLP models, seamlessly integrated with the Hugging Face ecosystem.
    What NLP tasks are available for training?
    Available today: binary classification, multi-class classification, regression, entity recognition, summarization, and automatic speech recognition.
    Which model languages are available?
    Available today: English, German, French, Spanish, Finnish, Swedish, Hindi, Dutch and more! Check out all supported languages in the documentation.
    How is my training data secure?
    Your training data stays on our server, and is private to your account. All data transfers are SSL-protected.
    How much does it cost?
    Based on how much training data and model variants are created, we send you a compute cost and payment link - as low as $10 per job.
    What training data formats are supported?
    CSV, TSV or JSON files, hosted anywhere. We delete your training data after training is done.
  • Create a new model

    Use AutoNLP CLI to create a model

  • huggingface@autonlp:~
    autonlp login --api-key YOUR_HUGGING_FACE_API_TOKEN
    autonlp create_project --name sentiment_detection --language en \
                           --task binary_classification --max_models 5
  • Upload your data

    We support csv and json file formats

  • huggingface@autonlp:~
    # Train split
    autonlp upload --project sentiment_detection --split train \
                   --col_mapping review:text,sentiment:target \
                   --files ~/datasets/train.csv
    # Validation split
    autonlp upload --project sentiment_detection --split valid \
                   --col_mapping review:text,sentiment:target \
                   --files ~/datasets/valid.csv
  • Run autonlp train

    AutoNLP automatically find the best model for your data

  • huggingface@autonlp:~
    autonlp train --project sentiment_detection
  • Track model progress

    Monitor and track model performance

  • huggingface@autonlp:~
    # Project progress
    autonlp project_info --name sentiment_detection
    # Project metrics
    autonlp metrics --project sentiment_detection
  • Use your model

    With its own HTTP endpoint or from the CLI

  • huggingface@autonlp:~
    autonlp predict --project sentiment_detection --model_id 42 --sentence "i love autonlp"