Hugging Face's logo

Reza2kn
/
visualears-fastconformer-fa-depoisoned-phaseB-streaming-mlx-fp16

Automatic Speech Recognition
MLX
Persian
visualears
fastconformer
streaming
fp16
rnnt
ctc
Model card Files
xet
 Community

VisualEars PhaseB Streaming MLX FP16

MLX FP16 weight bundle and RNNT runtime scaffold for Reza2kn/visualears-fastconformer-fa-depoisoned-phaseB.

Files:

  • phaseB_hybrid_streaming_fp16.safetensors
  • phaseB_hybrid_streaming_fp16_mlx_manifest.json
  • phaseb_mlx_rnnt_runtime.py
  • mlx_rnnt_runtime_smoke.json
  • mlx_vs_torch_rnnt_argmax20.json

Validation: MLX RNNT predictor/joint smoke passed; RNNT joint argmax matched TorchScript on 20 random probes.

The MLX encoder port is not complete in this artifact set. The exported weights and manifest preserve the streaming cache/head metadata for the full hybrid model.

Downloads last month

-

Downloads are not tracked for this model. How to track
MLX
Hardware compatibility
Log In to add your hardware

Quantized

Inference Providers NEW
Automatic Speech Recognition
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for Reza2kn/visualears-fastconformer-fa-depoisoned-phaseB-streaming-mlx-fp16

Base model

Reza2kn/visualears-fastconformer-fa-depoisoned-phaseB
Quantized
(6)
this model