MobileNetV3 Small 100
timm/mobilenetv3_small_100.lamb_in1k
published Dec 2022 · updated Oct 2025
MobileNetV3 Small 100 is a lightweight image classification model trained on ImageNet-1k using a LAMB optimizer recipe.
specs
| Task | Image Classification |
| Architecture | MobileNetV3 |
| Parameters | 2.5M |
| Input Size | 224 x 224 |
| Training Data | ImageNet-1k |
about this model
mobilenetv3_small_100.lamb_in1k is an image classification model that maps 224x224 RGB images to 1,000 ImageNet-1k class probabilities. It is a MobileNetV3-Small variant (2.5M parameters, 0.1 GMACs, 1.4M activations) trained with a
best for
- ·Classifying images into 1000 ImageNet categories
- ·Extracting feature maps for downstream computer vision tasks
- ·Generating image embeddings for similarity search or retrieval
FAQ
It is optimized for efficient image classification and feature extraction on resource-constrained devices, with a small parameter count.
It was trained using a LAMB optimizer recipe similar to ResNet Strikes Back A2, with EMA weight averaging, step LR schedule with warmup, and no CutMix.
Approximately 2.5 million parameters.
Input images should be resized to 224x224 pixels.
Use the OpenAI-compatible endpoint with your API key; refer to gigarouter documentation for details.
We're benchmarking and onboarding MobileNetV3 Small 100 as a hosted, OpenAI-compatible API. Sign in for free credit and be ready when it lands, or tell us you want it and we'll prioritize it.