recover_run_streaming.sh

./run_t5_mlm_flax_streaming.py \
    --output_dir="./" \
    --model_type="t5" \
    --config_name="./" \
    --tokenizer_name="./" \
    --model_name_or_path="./" \
    --dataset_name="pere/norwegian_colossal_corpus_v2_short100k" \
    --max_seq_length="512" \
    --weight_decay="0.01" \
    --per_device_train_batch_size="32" \
    --per_device_eval_batch_size="32" \
    --learning_rate="5.23e-3" \
    --warmup_steps="0" \
    --overwrite_output_dir \
    --cache_dir /mnt/disks/flaxdisk/cache/ \
    --num_train_epochs="5" \
    --adam_beta1="0.9" \
    --adam_beta2="0.98" \
    --logging_steps="500" \
    --num_train_steps="650000" \
    --num_eval_samples="5000" \
    --save_steps="5000" \
    --eval_steps="5000" \
    --preprocessing_num_workers 96 \
    --adafactor \
    --push_to_hub