maskit-learn 0.1.0

[MASK]it -Efficient pre-trained encoder adaption leveraging the [MASK]

[MASK]-It - Lightweight framework for efficient encoder model fine-tuning

Install

requires python 3.10 or above

pip install maskit-learn

Usage

Define task

# task and model
classes = ['happy', 'sad']
verbalizer_map = {'happy':['happy', 'fun'],
                  'sad':['sad', 'cry']}
model_name = 'google-bert/bert-base-uncased'

Select pre-trained model

from maskit.model import maskitModel
model = maskitModel(model_name=model_name,
                    verbalizer_map=verbalizer_map)

Load dataset

from torch.utils.data import DataLoader
from maskit.dataset import maskitDataset
text = ['I am so happy today that I cannot stay still','I am very very sad unfortunately']
labels = [1,0]
template = '{text}. This sentence is: [MASK]'
dataset = maskitDataset(text, labels, model_name, template)
dataloader = DataLoader(dataset=dataset, batch_size=2)

Inference

batch = next(iter(dataloader))
output = model(**batch)