From 56072f47cac0ce7c2e9b4bd0831d77f3232f8785 Mon Sep 17 00:00:00 2001 From: Mergen Nachin Date: Fri, 7 Nov 2025 15:20:55 -0500 Subject: [PATCH] Update README to include Whisper model link --- desktop/README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/desktop/README.md b/desktop/README.md index 5a76aeb5fcb..2c00be632e7 100644 --- a/desktop/README.md +++ b/desktop/README.md @@ -13,7 +13,7 @@ ExecuTorch is a lightweight, flexible runtime designed for efficient AI inferenc With increased demand for local inference on consumer desktops and laptops, exemplified by popular runtimes like llama.cpp and MLX, ExecuTorch is now experimenting with CUDA and Metal support. This is achieved by leveraging Inductor compiler technology from PyTorch, specifically using Ahead-of-Time Inductor [AOTI](https://docs.pytorch.org/docs/stable/torch.compiler_aot_inductor.html) to avoid reinventing the wheel. ## Key Benefits -- **Model Agnostic**: Validated on models such as [Voxtral](../examples/models/voxtral), [Gemma3-4b](../examples/models/gemma3), ResNet, and Whisper (WIP). Theoretically, any model exportable via torch.export is supported. +- **Model Agnostic**: Validated on models such as [Voxtral](../examples/models/voxtral), [Gemma3-4b](../examples/models/gemma3), ResNet, and [Whisper](../examples/models/whisper/README.md). Theoretically, any model exportable via torch.export is supported. - **PyTorch Ecosystem Integration**: Enables workflows for fine-tuning, quantization, and compilation within the PyTorch ecosystem. - **No Python Runtime During Inference**: Ideal for native applications (e.g., written in C++) embedding AI capabilities. - **No libtorch Dependency**: Reduces binary size, making deployment easier for resource-constrained applications.