patrickvonplaten
/

wavlm-libri-clean-100h-base-plus

Automatic Speech Recognition

librispeech_asr

Generated from Trainer

wavlm_libri_finetune

Inference Endpoints

Model card Files Files and versions Metrics Training metrics Community

patrickvonplaten commited on Dec 20, 2021

Commit

02c289c

•

1 Parent(s): b289767

Update run.sh

Files changed (1) hide show

run.sh +2 -2

run.sh CHANGED Viewed

@@ -2,11 +2,11 @@
 python -m torch.distributed.launch \
  	--nproc_per_node 8 run_speech_recognition_ctc.py \
 	--dataset_name="librispeech_asr" \
-	--model_name_or_path="microsoft/wavlm-base" \
 	--dataset_config_name="clean" \
 	--train_split_name="train.100" \
 	--eval_split_name="validation" \
-	--output_dir="./wavlm-libri-clean-100h-large" \
 	--preprocessing_num_workers="16" \
 	--length_column_name="input_length" \
 	--overwrite_output_dir \

 python -m torch.distributed.launch \
  	--nproc_per_node 8 run_speech_recognition_ctc.py \
 	--dataset_name="librispeech_asr" \
+	--model_name_or_path="microsoft/wavlm-base-plus" \
 	--dataset_config_name="clean" \
 	--train_split_name="train.100" \
 	--eval_split_name="validation" \
+	--output_dir="./wavlm-libri-clean-100h-base-plus" \
 	--preprocessing_num_workers="16" \
 	--length_column_name="input_length" \
 	--overwrite_output_dir \