TypeError: LlamaModel.__call__() got an unexpected keyword argument 'mask'

[laventura]

I'm getting a weird error, regardless of Python 3.12 or 3.13 on MacOS.

Wondering if folks have seen it, and/or how to resolve it??

Here's the error (with Python 3.12)

(mlx-audio-box) atul@BagPigMini:~/dev/mlx-audio-box|master⚡ ⇒  python -m mlx_audio.tts.generate --model Marvis-AI/marvis-tts-250m-v0.1 --text "Sample voice output from Marvis AI. Yay" --stream
Fetching 10 files: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10/10 [00:00<00:00, 285326.80it/s]

Model: Marvis-AI/marvis-tts-250m-v0.1
Text: Sample voice output from Marvis AI. Yay
Voice: None
Speed: 1.0x
Language: a
0it [00:00, ?it/s]
Error loading model: LlamaModel.__call__() got an unexpected keyword argument 'mask'
Traceback (most recent call last):
  File "/Users/atul/Dev/mlx-audio-box/.venv/lib/python3.12/site-packages/mlx_audio/tts/generate.py", line 302, in generate_audio
    for i, result in enumerate(results):
                     ^^^^^^^^^^^^^^^^^^
  File "/Users/atul/Dev/mlx-audio-box/.venv/lib/python3.12/site-packages/mlx_audio/tts/models/sesame/sesame.py", line 776, in generate
    sample = self.model.generate_frame(
             ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/atul/Dev/mlx-audio-box/.venv/lib/python3.12/site-packages/mlx_audio/tts/models/sesame/sesame.py", line 362, in generate_frame
    h = self.backbone(h, mask=curr_backbone_mask, cache=self.backbone_cache)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: LlamaModel.__call__() got an unexpected keyword argument 'mask'
(mlx-audio-box) atul@BagPigMini:~/dev/mlx-audio-box|master⚡ ⇒  python -V
Python 3.12.8

[lucataco]

I had to make the following edits to get it to run

[lucaorio]

I had success in working around this problem by downgrading some requirements:

"mlx-vlm<0.3.0",
"mlx-lm<0.28.0"