Merge branch 'main' of https://huggingface.co/marinone94/whisper-tiny-sv into main

run_speech_recognition_seq2seq_streaming.py

@@ -838,9 +838,10 @@ def main():
         kwargs["dataset_tags"] = dataset_names
         if data_args.dataset_train_config_name is not None:
             dataset_config_names = list(data_args.dataset_train_config_name.split(","))
-            kwargs["dataset"] = [f"{ds_name} {ds_cfg_name}" for ds_name, ds_cfg_name in zip(dataset_names, dataset_config_names)]
+            dataset_config_names_list = [f"{ds_name} {ds_cfg_name}" for ds_name, ds_cfg_name in zip(dataset_names, dataset_config_names)]
         else:
-            kwargs["dataset"] = dataset_names
+            dataset_config_names_list = dataset_names
+        kwargs["dataset"] = "\n".join(dataset_config_names_list)
         # if "common_voice" in data_args.dataset_name:
         #     kwargs["language"] = data_args.dataset_config_name[:2]
         if data_args.language_train is not None:

test_run_nordic.sh

@@ -2,11 +2,11 @@ python $1run_speech_recognition_seq2seq_streaming.py \
 	--model_name_or_path="openai/whisper-tiny" \
 	--dataset_train_name="mozilla-foundation/common_voice_11_0,mozilla-foundation/common_voice_11_0,mozilla-foundation/common_voice_11_0,babelbox/babelbox_voice,NbAiLab/NST,NbAiLab/NPSC,google/fleurs,google/fleurs,google/fleurs" \
 	--dataset_train_config_name="sv-SE,da,nn-NO,nst,no-distant,16K_mp3_nynorsk,sv_se,da_dk,nb_no" \
-	--language_train="swedish,danish,norwegian,swedish,norwegian,norwegian,swedish,danish,norwegian" \
+	--language_train="sv,da,no,sv,no,no,sv,da,no" \
 	--train_split_name="train+validation,train+validation,train+validation,train,train+test,train+validation,train+validation,train+validation,train+validation" \
 	--dataset_eval_name="mozilla-foundation/common_voice_11_0,mozilla-foundation/common_voice_11_0,mozilla-foundation/common_voice_11_0" \
 	--dataset_eval_config_name="sv-SE,da,nn-NO" \
-	--language_eval="swedish,danish,norwegian" \
+	--language_eval="sv,da,no" \
 	--eval_split_name="test" \
 	--model_index_name="Whisper Tiny Nordic" \
 	--max_train_samples="64" \