DistilBERT base model (uncased)

This is the uncased DistilBERT model fine-tuned on Multi-Genre Natural Language Inference (MNLI) dataset for the zero-shot classification task. The model is not case-sensitive, i.e., it does not make a difference between "english" and "English".

Training

Training is done on a p3.2xlarge AWS EC2 instance (1 NVIDIA Tesla V100 GPUs), with the following hyperparameters:

$ run_glue.py \
    --model_name_or_path distilbert-base-uncased \
    --task_name mnli \
    --do_train \
    --do_eval \
    --max_seq_length 128 \
    --per_device_train_batch_size 16 \
    --learning_rate 2e-5 \
    --num_train_epochs 5 \
    --output_dir /tmp/distilbert-base-uncased_mnli/

Evaluation results