Fix: Add missing torch import in generate_attention_visualization

Add attention visualization to Gradio demo with multi-modal display

Add attention visualization to Gradio demo with multi-modal display

Fix model architecture: decoder_embed_dim=384, add norm layer, correct forward method

Fix model architecture: decoder_embed_dim=384, add norm layer, correct forward method

Add GPU support via spaces module

Fix huggingface_hub compatibility issue, upgrade Gradio SDK to 5.29.0

Add online inference demo with 20 samples

Add arXiv paper link and citation (2605.13059)

Fix runtime: align sdk_version with Space default Gradio 6.14.0

Add requirements-train.txt for local training dependencies

Fix runtime: use empty Space requirements, move train deps to requirements-train.txt

Fix runtime error: remove gradio and huggingface_hub from requirements.txt

Import BrainAnytime code from GitHub and configure Gradio Space

initial commit