software-mansion/react-native-executorch-whisper-small.en

whisper-small.en

This repository hosts the whisper-small.en models exported for the React Native ExecuTorch library as ExecuTorch .pte programs, ready to run on device.

Upstream model: whisper-small.en

Variants

Path	Backend	Precision
coreml/whisper_small_en_coreml_fp16.pte	coreml	fp16
mlx/whisper_small_en_mlx_int8.pte	mlx	int8
vulkan/whisper_small_en_vulkan_fp16.pte	vulkan	fp16
vulkan/whisper_small_en_vulkan_int8.pte	vulkan	int8
xnnpack/whisper_small_en_xnnpack_fp32.pte	xnnpack	fp32
xnnpack/whisper_small_en_xnnpack_int8.pte	xnnpack	int8

XNNPACK int8

xnnpack/whisper_small_en_xnnpack_int8.pte is 8da8w (int8 dynamic activations, int8 per-channel weights). Greedy WER on LibriSpeech test-clean (150 clips): fp32 2.78, int8 2.72.

Repository structure

config.json                                38 B
coreml/config.json                         1.5 kB
coreml/whisper_small_en_coreml_fp16.pte    462 MB
mlx/config.json                            1.5 kB
mlx/whisper_small_en_mlx_int8.pte          276 MB
tokenizer.json                             2.3 MB
vulkan/config.json                         2.6 kB
vulkan/whisper_small_en_vulkan_fp16.pte    881 MB
vulkan/whisper_small_en_vulkan_int8.pte    484 MB
xnnpack/config.json                        2.6 kB
xnnpack/whisper_small_en_xnnpack_fp32.pte  1.1 GB
xnnpack/whisper_small_en_xnnpack_int8.pte  510 MB

Compatibility

These files are published for the ExecuTorch v1.4.1 runtime. ExecuTorch gives no forward compatibility guarantee, so an older runtime may fail to load them.

To use them in React Native ExecuTorch, pass the model constant shipped in the library's model registry to the corresponding task pipeline. See the documentation.

To load these files in your own ExecuTorch runtime, read the compatibility note first.