https://github.com/fastmachinelearning/qonnx_model_zoo
Model zoo for the Quantized ONNX (QONNX) model format
https://github.com/fastmachinelearning/qonnx_model_zoo
Last synced: 3 months ago
JSON representation
Model zoo for the Quantized ONNX (QONNX) model format
- Host: GitHub
- URL: https://github.com/fastmachinelearning/qonnx_model_zoo
- Owner: fastmachinelearning
- Created: 2021-09-15T16:45:36.000Z (over 4 years ago)
- Default Branch: main
- Last Pushed: 2025-02-09T14:22:25.000Z (about 1 year ago)
- Last Synced: 2025-02-09T15:28:14.220Z (about 1 year ago)
- Size: 22.5 MB
- Stars: 11
- Watchers: 14
- Forks: 3
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Contributing: CONTRIBUTING.md
Awesome Lists containing this project
README
# QONNX Model Zoo
This repo contains a variety of different, representative example models in the [Quantized ONNX](https://github.com/fastmachinelearning/qonnx) model format.
## Model Zoo Overview
| Model Name / Link | Dataset | Accuracy / Top-1 | NN Topology | Dominant Quantization | More Details |
|-------------------|-----------|------------------|-----------------|------------------------------------|--------------|
| [KWS MLP w3a3](https://github.com/fastmachinelearning/qonnx_model_zoo/tree/main/models/GSCV2) | GSCV2 | 87.89% | MLP | int3 weights / int3 activations | [README](https://github.com/fastmachinelearning/qonnx_model_zoo/blob/main/models/GSCV2/README.md) |
| [LFC_1W1A](https://github.com/fastmachinelearning/qonnx_model_zoo/tree/main/models/MNIST/Brevitas_FINN_LFC) | MNIST | 98.88% | MLP | bipolar weights / bipolar activations | [README](https://github.com/fastmachinelearning/qonnx_model_zoo/blob/main/models/MNIST/Brevitas_FINN_LFC/README.md) |
| [CNV_1W1A](https://github.com/fastmachinelearning/qonnx_model_zoo/tree/main/models/CIFAR10/Brevitas_FINN_CNV) | CIFAR-10 | 84.22% | VGG10-like | bipolar weights / bipolar activations | [README](https://github.com/fastmachinelearning/qonnx_model_zoo/tree/main/models/CIFAR10/Brevitas_FINN_CNV) |
| [CNV_1W1A](https://github.com/fastmachinelearning/qonnx_model_zoo/tree/main/models/GTSRB/Brevitas_CNV1W1A) | GTSRB | 96.93% | VGG10-like | bipolar weights / bipolar activations | [README](https://github.com/fastmachinelearning/qonnx_model_zoo/blob/main/models/GTSRB/Brevitas_CNV1W1A/README.md) |
| [mobilenet_4W4A](https://github.com/fastmachinelearning/qonnx_model_zoo/tree/main/models/ImageNet/Brevitas_FINN_mobilenet) | ImageNet | 71.14% | MobileNet-v1 | int4 weights / int4 activations | [README](https://github.com/fastmachinelearning/qonnx_model_zoo/blob/main/models/ImageNet/Brevitas_FINN_mobilenet/README.md) |
| [unsw_nb15_mlp_w2a2](https://github.com/fastmachinelearning/qonnx_model_zoo/tree/main/models/UNSW_NB15) | UNSW-NB15 | 91.90% | MLP | int2 weights / int2 activations | [README](https://github.com/fastmachinelearning/qonnx_model_zoo/blob/main/models/UNSW_NB15/README.md) |
| [quant_resnet18_w4a4_a2q_16b](https://github.com/fastmachinelearning/qonnx_model_zoo/blob/main/models/CIFAR10/a2q/README.md) | CIFAR10 | 94.2% | ResNet-18 | int4 weights / uint4 activations (int16 accumulators) | [README](https://github.com/fastmachinelearning/qonnx_model_zoo/blob/main/models/CIFAR10/a2q/README.md) |
| [ResNet-8](https://github.com/fastmachinelearning/qonnx_model_zoo/tree/main/models/CIFAR100) | CIFAR100 | 70.12% | ResNet-8 | int3 weights / int3 activations | [README](https://github.com/fastmachinelearning/qonnx_model_zoo/blob/main/models/CIFAR100/README.md) |
| [qkeras_jettagging](https://github.com/fastmachinelearning/qonnx_model_zoo/tree/main/models/JetTagging) | LHC jets | 76.2% | MLP | 6b weights / 6b activations | [README](https://github.com/fastmachinelearning/qonnx_model_zoo/blob/main/models/JetTagging/README.md) |
| [ResNet18 8w8a_e4m3](https://github.com/fastmachinelearning/qonnx_model_zoo/tree/main/models/CIFAR10/Brevitas_resnet18_float) | CIFAR10 | 93.09% | ResNet-18 | FP8 E4M3 | [README](https://github.com/fastmachinelearning/qonnx_model_zoo/blob/main/models/CIFAR10/Brevitas_resnet18_float/README.md) |
> **Note:** This table is incomplete. For a full list of models, see the [models directory](https://github.com/fastmachinelearning/qonnx_model_zoo/tree/main/models) on GitHub.