Fix doc example: mask_time_indices (numpy) has no attribute 'to' (#15033)

* fix doc example - AttributeError: 'numpy.ndarray' object has no attribute 'to'

* fix more

* Apply suggestions from code review

* Update src/transformers/models/unispeech/modeling_unispeech.py
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

src/transformers/models/unispeech/modeling_unispeech.py

@@ -1290,6 +1290,7 @@ class UniSpeechForPreTraining(UniSpeechPreTrainedModel):
         >>> batch_size, raw_sequence_length = input_values.shape
         >>> sequence_length = model._get_feat_extract_output_lengths(raw_sequence_length)
         >>> mask_time_indices = _compute_mask_indices((batch_size, sequence_length), mask_prob=0.2, mask_length=2)
+        >>> mask_time_indices = torch.tensor(mask_time_indices, device=input_values.device, dtype=torch.long)
 
         >>> with torch.no_grad():
         ...     outputs = model(input_values, mask_time_indices=mask_time_indices)

src/transformers/models/unispeech_sat/modeling_unispeech_sat.py

@@ -1322,6 +1322,7 @@ class UniSpeechSatForPreTraining(UniSpeechSatPreTrainedModel):
         >>> batch_size, raw_sequence_length = input_values.shape
         >>> sequence_length = model._get_feat_extract_output_lengths(raw_sequence_length)
         >>> mask_time_indices = _compute_mask_indices((batch_size, sequence_length), mask_prob=0.2, mask_length=2)
+        >>> mask_time_indices = torch.tensor(mask_time_indices, device=input_values.device, dtype=torch.long)
 
         >>> with torch.no_grad():
         ...     outputs = model(input_values, mask_time_indices=mask_time_indices)

src/transformers/models/wav2vec2/modeling_wav2vec2.py

@@ -1460,6 +1460,7 @@ class Wav2Vec2ForPreTraining(Wav2Vec2PreTrainedModel):
         >>> batch_size, raw_sequence_length = input_values.shape
         >>> sequence_length = model._get_feat_extract_output_lengths(raw_sequence_length)
         >>> mask_time_indices = _compute_mask_indices((batch_size, sequence_length), mask_prob=0.2, mask_length=2)
+        >>> mask_time_indices = torch.tensor(mask_time_indices, device=input_values.device, dtype=torch.long)
 
         >>> with torch.no_grad():
         ...     outputs = model(input_values, mask_time_indices=mask_time_indices)