diff --git a/egs/tedlium2/ASR/conformer_ctc2/.train.py.swp b/egs/tedlium2/ASR/conformer_ctc2/.train.py.swp
index 7e1995bf3..d054c650f 100644
Binary files a/egs/tedlium2/ASR/conformer_ctc2/.train.py.swp and b/egs/tedlium2/ASR/conformer_ctc2/.train.py.swp differ
diff --git a/egs/tedlium2/ASR/conformer_ctc2/train.py b/egs/tedlium2/ASR/conformer_ctc2/train.py
index 6c6149955..a3cf82b07 100755
--- a/egs/tedlium2/ASR/conformer_ctc2/train.py
+++ b/egs/tedlium2/ASR/conformer_ctc2/train.py
@@ -935,12 +935,14 @@ def run(rank, world_size, args):
     if params.print_diagnostics:
         diagnostic = diagnostics.attach_diagnostics(model)
 
-    librispeech = LibriSpeechAsrDataModule(args)
+    #librispeech = LibriSpeechAsrDataModule(args)
+    tedlium = TedAsrDataModule(args)
 
-    if params.full_libri:
-        train_cuts = librispeech.train_all_shuf_cuts()
-    else:
-        train_cuts = librispeech.train_clean_100_cuts()
+    #if params.full_libri:
+    #    train_cuts = librispeech.train_all_shuf_cuts()
+    #else:
+    #    train_cuts = librispeech.train_clean_100_cuts()
+    train_cuts = tedlium.train_cuts()
 
     def remove_short_and_long_utt(c: Cut):
         # Keep only utterances with duration between 1 second and 20 seconds
@@ -975,13 +977,20 @@ def run(rank, world_size, args):
     else:
         sampler_state_dict = None
 
-    train_dl = librispeech.train_dataloaders(
+    #train_dl = librispeech.train_dataloaders(
+    #    train_cuts, sampler_state_dict=sampler_state_dict
+    #)
+
+    #valid_cuts = librispeech.dev_clean_cuts()
+    #valid_cuts += librispeech.dev_other_cuts()
+    #valid_dl = librispeech.valid_dataloaders(valid_cuts)
+    
+    train_dl = tedlium.train_dataloaders(
         train_cuts, sampler_state_dict=sampler_state_dict
     )
 
-    valid_cuts = librispeech.dev_clean_cuts()
-    valid_cuts += librispeech.dev_other_cuts()
-    valid_dl = librispeech.valid_dataloaders(valid_cuts)
+    valid_cuts = tedlium.dev_cuts()
+    valid_dl = tedlium.valid_dataloaders(valid_cuts)
 
     if params.print_diagnostics:
         scan_pessimistic_batches_for_oom(