Views
No views yet
| Source | OpenMOSS-Team/MOSS-Transcribe-preview-2B @ c98175cb20e48bd9be4e95f6c85f2af18899f780 (bf16, 4.84 GB) |
| Architecture | Qwen3-Omni-MoE audio encoder (32 layers) + gated-MLP adapter + Qwen3-1.7B decoder (28 layers) |
| Format | MLX, single safetensors tree |
| Quantization | 4-bit affine, group-size 64 (Qwen3 decoder only, 197 tensors; audio encoder + adapter preserved bf16) |
| Disk size | 2.2 GB |
model.audio_model.* (525 tensors) and model.audio_adapter.* (3 tensors) remain BF16, while only model.language_model.* is quantized.e1.wav, e2.wav, and e3.wav. non-English audio is out of scope for this preview card copy and does not participate in the release verdict.| clip | CER vs MLX-BF16 | gating | wall_s |
|---|---|---|---|
e1.wav | 0.0 | yes | 0.109 |
e2.wav | 0.0 | yes | 0.113 |
e3.wav | 0.0 | yes | 0.126 |
majentik/MOSS-Transcribe-preview-2B-MLX-4bit (this repo)majek/pipelines/moss_transcribe_mlx (Qwen3-Omni audio encoder vendored from mlx-vlm, decoder via stock mlx_lm.models.qwen3):1python -m pipelines.moss_transcribe_mlx.generate \
2 --path majentik/MOSS-Transcribe-preview-2B-MLX-4bit \
3 --wav your_clip.wavmlx-lm/mlx-vlm alone cannot load this composite audio architecture. This is an English ASR model (Open ASR Leaderboard); non-English audio is out of domain.1python -m pipelines.mlx_direct_quantize --model moss-transcribe-preview-2b \
2 --base-dir <upstream snapshot> --out-dir <out> \
3 --bits 4 --mode affine --group-size 64