export LD_LIBRARY_PATH=/usr/lib/x86_64-linux-gnu
python run_speech_recognition_seq2seq_streaming.py \
	--model_name_or_path="openai/whisper-small" \
	--train_dataset_names "/home/ubuntu/train_asr_212k.csv" \
	--train_dataset_config_names "" \
    --train_dataset_split_names "train" \
    --train_dataset_text_column_names "sentence" \
    --eval_dataset_name="mozilla-foundation/common_voice_11_0" \
    --eval_dataset_config_name="fi" \
    --eval_split_name="test" \
	--language="finnish" \
	--model_index_name="Whisper Small Finnish" \
	--max_steps="10000" \
	--output_dir="./outputs" \
	--per_device_train_batch_size="32" \
	--per_device_eval_batch_size="16" \
	--logging_steps="25" \
	--learning_rate="1e-5" \
	--warmup_steps="500" \
	--evaluation_strategy="steps" \
	--eval_steps="1000" \
	--save_strategy="steps" \
	--save_steps="1000" \
	--generation_max_length="225" \
	--length_column_name="input_length" \
	--max_duration_in_seconds="30" \
	--text_column_name="sentence" \
	--freeze_feature_encoder="False" \
	--report_to="tensorboard" \
	--metric_for_best_model="wer" \
	--greater_is_better="False" \
    --load_best_model_at_end \
    --optim="adamw_bnb_8bit" \
	--gradient_checkpointing \
	--fp16 \
	--overwrite_output_dir \
	--do_train \
	--do_eval \
	--predict_with_generate \
	--do_normalize_eval \
	--use_auth_token \
	--push_to_hub