From 93c26434c8c2e65355926ec03d32f58187286eb3 Mon Sep 17 00:00:00 2001 From: Mergen Nachin Date: Mon, 9 Mar 2026 17:40:57 -0400 Subject: [PATCH 1/7] Update README with Mistral pre-exported .pte files info Added a tip about pre-exported .pte files for Mistral. --- examples/models/voxtral_realtime/README.md | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/examples/models/voxtral_realtime/README.md b/examples/models/voxtral_realtime/README.md index 7d29ba8c11b..bed6c27916e 100644 --- a/examples/models/voxtral_realtime/README.md +++ b/examples/models/voxtral_realtime/README.md @@ -58,6 +58,10 @@ python -m executorch.extension.audio.mel_spectrogram \ Export produces a single `.pte` containing the audio encoder, text decoder, and token embedding. +> [!Tip] +> Mistral has already published pre-exported .pte files for select backends, including macOS Metal, on their [Hugging Face Hub](https://huggingface.co/mistral-labs/Voxtral-Mini-4B-Realtime-2602-Executorch). + + ```bash python export_voxtral_rt.py \ --model-path ~/models/Voxtral-Mini-4B-Realtime-2602 \ From c4694053ca60edfd1b4cd063a3461d40827d20fb Mon Sep 17 00:00:00 2001 From: Mergen Nachin Date: Mon, 9 Mar 2026 18:17:22 -0400 Subject: [PATCH 2/7] Update examples/models/voxtral_realtime/README.md Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> --- examples/models/voxtral_realtime/README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/examples/models/voxtral_realtime/README.md b/examples/models/voxtral_realtime/README.md index bed6c27916e..ddce6421026 100644 --- a/examples/models/voxtral_realtime/README.md +++ b/examples/models/voxtral_realtime/README.md @@ -59,7 +59,7 @@ Export produces a single `.pte` containing the audio encoder, text decoder, and token embedding. > [!Tip] -> Mistral has already published pre-exported .pte files for select backends, including macOS Metal, on their [Hugging Face Hub](https://huggingface.co/mistral-labs/Voxtral-Mini-4B-Realtime-2602-Executorch). +> Mistral has already published pre-exported `.pte` files for select backends, including macOS Metal, on their [Hugging Face Hub](https://huggingface.co/mistral-labs/Voxtral-Mini-4B-Realtime-2602-Executorch). ```bash From 51ccf97948d212330b4e6cf65c934bbd4012cbe6 Mon Sep 17 00:00:00 2001 From: Mergen Nachin Date: Mon, 9 Mar 2026 22:08:28 -0400 Subject: [PATCH 3/7] Update README with macOS app link for transcription Added links to a standalone macOS app for real-time transcription. --- examples/models/voxtral_realtime/README.md | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/examples/models/voxtral_realtime/README.md b/examples/models/voxtral_realtime/README.md index ddce6421026..803305cd81a 100644 --- a/examples/models/voxtral_realtime/README.md +++ b/examples/models/voxtral_realtime/README.md @@ -22,6 +22,10 @@ microphone input). https://github.com/user-attachments/assets/44717dc5-777f-4710-ad55-5ec4fa04b9c4 +Also, try a sample [standalone macOS app](https://github.com/meta-pytorch/executorch-examples/tree/main/voxtral_realtime/macos) to do realtime transcription + +https://github.com/user-attachments/assets/0ce4fb05-76b2-47e4-85e2-6f49b4e9e34e + ## Prerequisites - ExecuTorch installed from source (see [building from source](../../../docs/source/using-executorch-building-from-source.md)) From 22aaf881e6f50b6cfe015f1f4627f26627b3de76 Mon Sep 17 00:00:00 2001 From: Mergen Nachin Date: Tue, 10 Mar 2026 07:53:14 -0400 Subject: [PATCH 4/7] Update examples/models/voxtral_realtime/README.md Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> --- examples/models/voxtral_realtime/README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/examples/models/voxtral_realtime/README.md b/examples/models/voxtral_realtime/README.md index 803305cd81a..d31f7363a32 100644 --- a/examples/models/voxtral_realtime/README.md +++ b/examples/models/voxtral_realtime/README.md @@ -63,7 +63,7 @@ Export produces a single `.pte` containing the audio encoder, text decoder, and token embedding. > [!Tip] -> Mistral has already published pre-exported `.pte` files for select backends, including macOS Metal, on their [Hugging Face Hub](https://huggingface.co/mistral-labs/Voxtral-Mini-4B-Realtime-2602-Executorch). +> Mistral has already published pre-exported `.pte` files for select backends, including macOS Metal, on their [HuggingFace Hub](https://huggingface.co/mistral-labs/Voxtral-Mini-4B-Realtime-2602-Executorch). ```bash From 4ca3c7e6cafc616e8d2a81358ca5c227fcf31bfd Mon Sep 17 00:00:00 2001 From: Mergen Nachin Date: Tue, 10 Mar 2026 07:53:32 -0400 Subject: [PATCH 5/7] Apply suggestion from @Copilot Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> --- examples/models/voxtral_realtime/README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/examples/models/voxtral_realtime/README.md b/examples/models/voxtral_realtime/README.md index d31f7363a32..3d80089d35b 100644 --- a/examples/models/voxtral_realtime/README.md +++ b/examples/models/voxtral_realtime/README.md @@ -22,7 +22,7 @@ microphone input). https://github.com/user-attachments/assets/44717dc5-777f-4710-ad55-5ec4fa04b9c4 -Also, try a sample [standalone macOS app](https://github.com/meta-pytorch/executorch-examples/tree/main/voxtral_realtime/macos) to do realtime transcription +Also, try a sample [standalone macOS app](https://github.com/meta-pytorch/executorch-examples/tree/main/voxtral_realtime/macos) to do real time transcription https://github.com/user-attachments/assets/0ce4fb05-76b2-47e4-85e2-6f49b4e9e34e From bc7ac2899d53383a58f12ca86b9d1091abb6d97e Mon Sep 17 00:00:00 2001 From: Mergen Nachin Date: Tue, 10 Mar 2026 07:57:36 -0400 Subject: [PATCH 6/7] Fix formatting and update links in README.md --- examples/models/voxtral_realtime/README.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/examples/models/voxtral_realtime/README.md b/examples/models/voxtral_realtime/README.md index 3d80089d35b..85e95d0e66b 100644 --- a/examples/models/voxtral_realtime/README.md +++ b/examples/models/voxtral_realtime/README.md @@ -22,9 +22,9 @@ microphone input). https://github.com/user-attachments/assets/44717dc5-777f-4710-ad55-5ec4fa04b9c4 -Also, try a sample [standalone macOS app](https://github.com/meta-pytorch/executorch-examples/tree/main/voxtral_realtime/macos) to do real time transcription +Also, try a sample [standalone macOS app](https://github.com/meta-pytorch/executorch-examples/tree/main/voxtral_realtime/macos) to do real time transcription. -https://github.com/user-attachments/assets/0ce4fb05-76b2-47e4-85e2-6f49b4e9e34e +https://github.com/user-attachments/assets/a89677ef-8309-426e-a2f4-6d1ea8494b99 ## Prerequisites From d55ab1b6a01128bd8c2379707fb07564b2b681f2 Mon Sep 17 00:00:00 2001 From: Mergen Nachin Date: Tue, 10 Mar 2026 08:02:01 -0400 Subject: [PATCH 7/7] Update examples/models/voxtral_realtime/README.md Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> --- examples/models/voxtral_realtime/README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/examples/models/voxtral_realtime/README.md b/examples/models/voxtral_realtime/README.md index 85e95d0e66b..ab6dcf3a6c9 100644 --- a/examples/models/voxtral_realtime/README.md +++ b/examples/models/voxtral_realtime/README.md @@ -62,7 +62,7 @@ python -m executorch.extension.audio.mel_spectrogram \ Export produces a single `.pte` containing the audio encoder, text decoder, and token embedding. -> [!Tip] +> [!TIP] > Mistral has already published pre-exported `.pte` files for select backends, including macOS Metal, on their [HuggingFace Hub](https://huggingface.co/mistral-labs/Voxtral-Mini-4B-Realtime-2602-Executorch).