error in encoder thread narrow invalid args start + len > dim_len: [4096, 32], dim: 0, start: 4096, len:2

[RoversX]

Backend impacted

The MLX implementation

Operating system

Mac OS X

Hardware

Metal with MLX

Description

When running python -m moshi_mlx.local_web -q 4

(moshi) moshi % python -m moshi_mlx.local_web -q 4
[Info] [SERVER] loading text tokenizer /Users/.cache/huggingface/hub/models--kyutai--moshiko-mlx-q4/snapshots/18e4df760a34dd5977a34517d7d1580e07acbb2f1/tokenizer_spm_32k_3.model
[Info] [SERVER] loading weights /Users/.cache/huggingface/hub/models--kyutai--moshiko-mlx-q4/snapshots/18e4df760a34dd5977a34517d7d1580e07acbb2f1/model.q4.safetensors
[Info] [SERVER] weights loaded
[Info] [SERVER] model warmed up
[Info] [SERVER] connected!
[Info] [CLIENT] received 'start' from server, starting...
[Info] retrieving the static content
[Info] serving static content from /Users/.cache/huggingface/hub/models--kyutai--moshi-artifacts/snapshots/8481e95f73827e4e70ac7311c12b0be099276182/dist
[Info] listening to http://localhost:8998
[Info] opening browser at http://localhost:8998
[Info] accepted connection
error in encoder thread narrow invalid args start + len > dim_len: [4096, 32], dim: 0, start: 4096, len:2
error in encoder thread narrow invalid args start + len > dim_len: [4096, 32], dim: 0, start: 4096, len:2
error in encoder thread narrow invalid args start + len > dim_len: [4096, 32], dim: 0, start: 4096, len:2
error in encoder thread narrow invalid args start + len > dim_len: [4096, 32], dim: 0, start: 4096, len:2
error in encoder thread narrow invalid args start + len > dim_len: [4096, 32], dim: 0, start: 4096, len:2
error in encoder thread narrow invalid args start + len > dim_len: [4096, 32], dim: 0, start: 4096, len:2
| [LAG]error in encoder thread narrow invalid args start + len > dim_len: [4096, 32], dim: 0, start: 4096, len:2
[LAG]error in encoder thread narrow invalid args start + len > dim_len: [4096, 32], dim: 0, start: 4096, len:2
[LAG]error in encoder thread narrow invalid args start + len > dim_len: [4096, 32], dim: 0, start: 4096, len:2

Extra information

$ pip install moshi_mlx
$ python -m moshi_mlx.local_web -q 4

Environment

System Information:

• OS: macOS 15.0 (24A335) arm64
• Python Version: Python 3.12.6 (main, Sep 6 2024, 19:03:47) [Clang 15.0.0 (clang-1500.3.9.4)] on darwin
• Host: MacBook Pro 18,3
• Kernel: 24.0.0
• Uptime: 1 day, 4 hours, 7 mins
• Packages: 131 (installed via Homebrew)
• Shell: zsh 5.9
• Resolution: 1352x878 @2x
• Desktop Environment (DE): Aqua
• Window Manager (WM): Quartz Compositor
• Window Theme: Blue (Dark)
• Terminal: Apple Terminal
• Terminal Font: SFMono-Regular

Hardware Information:

• CPU: Apple M1 Pro
• GPU: Apple M1 Pro
• Memory: 1626MiB / 16384MiB

[RoversX]

reaches the max cache size: #51