Add model weights

.gitattributes

@@ -30,3 +30,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.whisper filter=lfs diff=lfs merge=lfs -text

medium.en.whisper

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e82e4a62220c0dc0aae204900c6061708339307e82694ad61ccd7fcb922193a8
+size 3055771163

run_librispeech.sh

@@ -0,0 +1,38 @@
+#!/usr/bin/env bash
+CUDA_VISIBLE_DEVICES=0 python run_speech_recognition_whisper.py \
+	--model_name_or_path="medium.en" \
+	--dataset_name="librispeech_asr" \
+	--dataset_config_name="all" \
+	--train_split_name="train.clean.100+train.clean.360+train.other.500" \
+	--eval_split_name="validation.clean" \
+	--test_split_name="validation.other+test.clean+test.other" \
+	--max_steps="5000" \
+	--text_column_name="text" \
+	--output_dir="./" \
+	--run_name="whisper-ls-960h-5k" \
+	--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="1000" \
+	--save_strategy="steps" \
+	--save_steps="1000" \
+	--generation_max_length="224" \
+	--length_column_name="input_lengths" \
+	--do_lower_case="True" \
+	--push_to_hub="False" \
+	--gradient_checkpointing \
+	--group_by_length \
+	--freeze_encoder \
+	--fp16 \
+	--overwrite_output_dir \
+	--do_train \
+	--do_eval \
+	--do_predict \
+	--predict_with_generate \
+	--use_auth_token