| --- |
| license: mit |
| base_model: Ruicheng/moge-2-vitl-normal |
| tags: |
| - webgpu |
| - depth-estimation |
| - surface-normals |
| - monocular-depth |
| pipeline_tag: depth-estimation |
| --- |
| |
| # MoGe-2 WebGPU weights |
|
|
| Pre-converted fp16 weights for [moge-webgpu](https://github.com/lyonsno/moge-webgpu) — a complete port of [MoGe-2](https://github.com/microsoft/MoGe) (DINOv2 ViT-Large + ConvStack decoder, `Ruicheng/moge-2-vitl-normal`) from PyTorch to pure WebGPU compute shaders. No server, no WASM, no ONNX runtime. |
|
|
| ## Files |
|
|
| - `weights.bin` (~660MB) — flat fp16 binary, all model tensors concatenated for direct WebGPU buffer upload |
| - `weights.json` — tensor manifest (names, shapes, offsets) |
|
|
| ## Usage |
|
|
| The [moge-webgpu](https://github.com/lyonsno/moge-webgpu) app streams these weights automatically on first load. To reproduce the conversion from the original checkpoint: |
|
|
| ```bash |
| python tools/convert_weights.py \ |
| --model Ruicheng/moge-2-vitl-normal \ |
| --output public/weights.bin \ |
| --dtype fp16 |
| ``` |
|
|
| ## License |
|
|
| MIT, matching upstream MoGe-2. Original model by Microsoft Research ([MoGe-2 paper](https://arxiv.org/abs/2507.02546)). |
|
|
