return_dict in encodec is always set to True:

[kamilakesbi]

System Info

• transformers version: 4.42.0.dev0
• Platform: Linux-5.4.0-166-generic-x86_64-with-glibc2.29
• Python version: 3.8.10
• Huggingface_hub version: 0.23.3
• Safetensors version: 0.4.2
• Accelerate version: 0.29.1
• Accelerate config: not found
• PyTorch version (GPU?): 2.2.2+cu121 (True)
• Tensorflow version (GPU?): 2.13.1 (True)
• Flax version (CPU?/GPU?/TPU?): 0.7.0 (cpu)
• Jax version: 0.4.13
• JaxLib version: 0.4.13
• Using distributed or parallel set-up in script?:
• Using GPU in script?:
• GPU type: NVIDIA DGX Display

Reproduction

from transformers import EncodecConfig, EncodecModel, EncodecFeatureExtractor
import numpy as np 
import torch

signal = [np.random.randn(1065473), np.random.randn(1065473) ]
signal = EncodecFeatureExtractor()(signal)

model = EncodecModel(EncodecConfig())

result = model(torch.tensor(signal.input_values), return_dict = True)
print(type(result))
result2 = model(torch.tensor(signal.input_values), return_dict = False)
print(type(result2))

will print <class 'transformers.models.encodec.modeling_encodec.EncodecOutput'> even when return_dict is set to False.

Expected behavior

It should return a tuple when return_dict is set to False.

cc @sanchit-gandhi

[AbhinavJangra29]

transformers/src/transformers/models/encodec/modeling_encodec.py

Line 789 in dfaadfd

return_dict = return_dict or self.config.return_dict

@kamilakesbi
by default self.config.return_dict is true so the or condition is always maintained and the function returns a dict.

[kamilakesbi]

Hi @AbhinavJangra29,

Thanks for spotting the issue. @jla524 has opened a PR to fix it!