From e964ff20e56fccfa7718ecba04d18f0434100cad Mon Sep 17 00:00:00 2001 From: cianmcnally Date: Wed, 15 Jul 2026 13:17:37 +0200 Subject: [PATCH] Update AutoGaze model ID to nvidia/AutoGaze in QUICK_START.md MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The bfshi/AutoGaze model repository no longer exists on Hugging Face — the model now lives at nvidia/AutoGaze, so the from_pretrained calls in the quick-start examples fail with a 404. --- QUICK_START.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/QUICK_START.md b/QUICK_START.md index 35ecf42..631d4bb 100644 --- a/QUICK_START.md +++ b/QUICK_START.md @@ -13,8 +13,8 @@ from autogaze.datasets.video_utils import read_video_pyav, transform_video_for_p from autogaze.models.autogaze import AutoGazeImageProcessor, AutoGaze # Load AutoGaze model and its preprocessor from HuggingFace -autogaze_transform = AutoGazeImageProcessor.from_pretrained("bfshi/AutoGaze") -autogaze_model = AutoGaze.from_pretrained("bfshi/AutoGaze") +autogaze_transform = AutoGazeImageProcessor.from_pretrained("nvidia/AutoGaze") +autogaze_model = AutoGaze.from_pretrained("nvidia/AutoGaze") # Load a video video_path = "/assets/example_input.mp4" # Fill in the path of AutoGaze codebase @@ -92,7 +92,7 @@ Let's take SigLIP2-SO400M with 384x384 input resolution and 14x14 patch size as ```python # Load AutoGaze preprocessor with size of 392 -autogaze_392_transform = AutoGazeImageProcessor.from_pretrained("bfshi/AutoGaze", size=(392, 392)) +autogaze_392_transform = AutoGazeImageProcessor.from_pretrained("nvidia/AutoGaze", size=(392, 392)) # Preprocess the video with AutoGaze preprocessor video_input_autogaze_392 = transform_video_for_pytorch(raw_video, autogaze_392_transform) # T * C * H * W