A small package for convenient predictions on fine-tuned huggingface-models
Project description
This package is meant to offer a variety of convenient inference functions, particularly for fine-tuned transformers.
Limited to Jochen Hartmann's fine-tuned j-hartmann/emotion-english-distilroberta-base for now.
Install:
- Command line
pip install ezpredict
- Google Colab
!pip install ezpredict
Imports:
pip install numpy
pip install torch
pip install transformers
How to use:
from ezpredict import predict
preds = predict.predict_input(model_name="j-hartmann/emotion-english-distilroberta-base",
input=["What a beautiful day!"],
return_values=True,
print_values=True)
Variables:
model_name
-> name of model to perform inference on (limited to Jochen Hartmann's fine-tuned j-hartmann/emotion-english-distilroberta-base for now)input
-> list of strings to perform predictions onreturn_values
-> True/False, True: returns predictions as list of tuplesprint_values
-> True/False, True: returns verbose outputs
Sample output:
[('What a beautiful day!', {'anger': 0.0013013791, 'disgust': 0.00047031444, 'fear': 0.001256481, 'joy': 0.95772415, 'neutral': 0.005870249, 'sadness': 0.004233605, 'surprise': 0.029143812}), ('My grandfather died today.', {'anger': 0.0012322478, 'disgust': 0.002399753, 'fear': 0.0018630251, 'joy': 0.0021994421, 'neutral': 0.012114782, 'sadness': 0.96717805, 'surprise': 0.013012686})]
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
ezpredict-0.0.6.tar.gz
(3.7 kB
view hashes)
Built Distribution
Close
Hashes for ezpredict-0.0.6-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | a1681efeb706fc020aa0fece5e92a194fc5a125d4ea14b5e4358175912039de9 |
|
MD5 | 30a85ceb4cba6db37367e37aec3f1bb6 |
|
BLAKE2b-256 | edc16449f78065b17d2aec9ec97caf875b4e6d8b889d0d2279fbc7a7854c1823 |