distilbert-base-uncased finetuned on MNLI

Model Details and Training Data

We used the pretrained model from distilbert-base-uncased and finetuned it on MultiNLI dataset.

The training parameters were kept the same as Devlin et al., 2019 (learning rate = 2e-5, training epochs = 3, max_sequence_len = 128 and batch_size = 32).

Evaluation Results

The evaluation results are mentioned in the table below.

Test Corpus Accuracy
Matched 0.8223
Mismatched 0.8216
Downloads last month
65
Hosted inference API
Text Classification
This model can be loaded on the Inference API on-demand.