| library_name: pytorch | |
| pipeline_tag: image-classification | |
| license: mit | |
| tags: | |
| - pytorch | |
| - torchscript | |
| - image-classification | |
| - tiny | |
| - smoke-test | |
| # Tiny TorchScript CNN | |
| Minimal CNN traced to TorchScript for pipeline validation and testing. | |
| ## Model Details | |
| - **Architecture**: Simple 3-layer CNN (TinyCNN) | |
| - **Input**: `[1, 3, 224, 224]` float32 tensor (RGB image) | |
| - **Output**: `[1, 10]` logits (10 classes) | |
| - **Format**: TorchScript (traced) | |
| ## Usage | |
| ```python | |
| import torch | |
| from huggingface_hub import hf_hub_download | |
| # Download and load | |
| model_path = hf_hub_download("Khauneesh/tiny-cnn-torchscript", "model.pt") | |
| model = torch.jit.load(model_path, map_location="cpu") | |
| model.eval() | |
| # Inference | |
| x = torch.randn(1, 3, 224, 224) | |
| with torch.no_grad(): | |
| logits = model(x) | |
| ``` | |
| ## Testing | |
| This model is designed for smoke testing TorchScript deployment pipelines. | |