esb/whisper-aed-tedlium

To reproduce this run, first install Whisper from the Transformers compatible repo patrickvonplaten/whisper:

pip install git+https://github.com/openai/whisper.git

Then execute the command:

#!/usr/bin/env bash
CUDA_VISIBLE_DEVICES=0 python run_speech_recognition_whisper.py \
    --model_name_or_path="medium.en" \
  --dataset_name="esb/datasets" \
  --dataset_config_name="tedlium" \
    --max_steps="2500" \
    --output_dir="./" \
    --run_name="whisper-tedlium" \
    --wandb_project="whisper" \
    --per_device_train_batch_size="64" \
    --per_device_eval_batch_size="16" \
    --logging_steps="25" \
    --learning_rate="1e-4" \
    --warmup_steps="500" \
    --report_to="wandb" \
    --preprocessing_num_workers="16" \
    --evaluation_strategy="steps" \
    --eval_steps="500" \
    --save_strategy="steps" \
    --save_steps="500" \
    --generation_max_length="224" \
    --length_column_name="input_lengths" \
    --gradient_checkpointing \
    --group_by_length \
    --freeze_encoder \
    --fp16 \
    --overwrite_output_dir \
    --do_train \
    --do_eval \
    --do_predict \
    --predict_with_generate \
    --use_auth_token