./run_mlm_flax_stream.py \
    --output_dir="./" \
    --model_type="roberta" \
    --config_name="./" \
    --tokenizer_name="./" \
    --model_name_or_path="./" \
    --dataset_name="NbAiLab/scandinavian" \
    --max_seq_length="128" \
    --weight_decay="0.01" \
    --per_device_train_batch_size="128" \
    --per_device_eval_batch_size="128" \
    --learning_rate="6e-5" \
    --warmup_steps="5000" \
    --overwrite_output_dir \
    --cache_dir /mnt/disks/flaxdisk/cache/ \
    --num_train_steps="180000" \
    --adam_beta1="0.9" \
    --adam_beta2="0.98" \
    --logging_steps="10000" \
    --save_steps="10000" \
    --eval_steps="10000" \
    --preprocessing_num_workers 96 \
    --auth_token True \
    --adafactor \
    --push_to_hub

Train for 20k steps for 512 sequences:

bash

./run_mlm_flax_stream.py \
    --output_dir="./" \
    --model_type="roberta" \
    --config_name="./" \
    --tokenizer_name="./" \
    --model_name_or_path="./" \
    --dataset_name="NbAiLab/scandinavian" \
    --max_seq_length="512" \
    --weight_decay="0.01" \
    --per_device_train_batch_size="48" \
    --per_device_eval_batch_size="48" \
    --learning_rate="3e-5" \
    --warmup_steps="5000" \
    --overwrite_output_dir \
    --cache_dir /mnt/disks/flaxdisk/cache/ \
    --num_train_steps="20000" \
    --adam_beta1="0.9" \
    --adam_beta2="0.98" \
    --logging_steps="20000" \
    --save_steps="10000" \
    --eval_steps="10000" \
    --preprocessing_num_workers 96 \
    --auth_token True \
    --adafactor \
    --push_to_hub

Approximate additional training time: 1 week.

pere/nb-roberta-base-scandinavian-long

This is just a Test Model. Do NOT use for anything!

Train for 180k steps for 128 sequences:

Train for 20k steps for 512 sequences:

pere/
nb-roberta-base-scandinavian-long