#!/usr/bin/env bash
python3 run_mlm_flax.py \
    --model_name_or_path="flax_model.msgpack" \
    --output_dir="./" \
    --model_type="roberta" \
    --config_name="./" \
    --tokenizer_name="./" \
    --dataset_name="oscar" \
    --dataset_config_name="unshuffled_deduplicated_id" \
    --max_seq_length="128" \
    --preprocessing_num_workers="64" \
    --per_device_train_batch_size="128" \
    --per_device_eval_batch_size="128" \
    --learning_rate="2e-4" \
    --warmup_steps="1000" \
    --overwrite_output_dir \
    --num_train_epochs="8" \
    --push_to_hub