To train, fine-tune or play with the model you will need to install NVIDIA NeMo. We recommend you install it after you've installed latest Pytorch version.

pip install nemo_toolkit['all']

How to Use this Model

The model is available for use in the NeMo toolkit [3], and can be used as a pre-trained checkpoint for inference or for fine-tuning on another dataset.

Automatically instantiate the model

python

import nemo.collections.nlp as model
model = nemo_nlp.models.question_answering.qa_bert_model.BERTQAModel.from_pretrained("raihanpf22/nlp_id_qa_bert_base_uncased")

Transcribing using Python

Simply do:

eval_device = [config.trainer.devices[0]] if isinstance(config.trainer.devices, list) else 1
model.trainer = pl.Trainer(
    devices=eval_device,
    accelerator=config.trainer.accelerator,
    precision=16,
    logger=False,
)

config.exp_manager.create_checkpoint_callback = False
exp_dir = exp_manager(model.trainer, config.exp_manager)
output_nbest_file = os.path.join(exp_dir, "output_nbest_file.json")
output_prediction_file = os.path.join(exp_dir, "output_prediction_file.json")

all_preds, all_nbest = model.inference(
    "questions.json",
    output_prediction_file=output_prediction_file,
    output_nbest_file=output_nbest_file,
    num_samples=-1, # setting to -1 will use all samples for inference
)

for question_id in all_preds:
    print(all_preds[question_id])

Input

This model accepts SQuAD Format v2.0 as input.

Output

This model provides output in the form of answers to questions according to the existing context.

Model Architecture

Using an uncased BERT base architecture model.

Training

50 Epochs, 8 Batch size per GPU, 1 num_layer

Datasets

using SQuAD v2.0 as train data

Performance

test_HasAns_exact = 98.0 test_HasAns_f1 = 98.0465087890625 test_HasAns_total = 100.0 test_NoAns_exact = 0.0 test_NoAns_f1 = 0.0 test_NoAns_total = 0.0 test_exact = 98.0 test_f1 = 98.0465087890625 test_loss = 0.00019806227646768093 test_total = 100.0

References

[1] NVIDIA NeMo Toolkit