# HardCoRe-NAS

**Category:** Efficient/Mobile · **Code:**
[`timm/models/hardcorenas.py`](https://github.com/huggingface/pytorch-image-models/blob/main/timm/models/hardcorenas.py)
· **Architectures:** 6 · **Pretrained weights:** 6 (6 with ImageNet-1k results)

Mobile networks found by HardCoRe-NAS, a differentiable neural architecture search with hard latency constraints, built
on the `MobileNetV3` class.

## Papers

* [HardCoRe-NAS: Hard Constrained diffeRentiable Neural Architecture Search](https://arxiv.org/abs/2102.11646)

## Usage

See the [Quickstart](../quickstart) for loading pretrained weights, image preprocessing and fine-tuning, and [Feature
Extraction](../feature_extraction) for using the models as backbones.

```py
>>> import timm
>>> model = timm.create_model('hardcorenas_f.miil_green_in1k', pretrained=True)
>>> data_config = timm.data.resolve_model_data_config(model)
```

## Model API support

| Feature | `MobileNetV3` |
|---|---|
| [`features_only=True` feature maps](../model_api#intermediate-features) | ✓ (strides 2, 4, 8, 16, 32) |
| [`forward_intermediates()`](../model_api#intermediate-features) | ✓ |
| [`prune_intermediate_layers()`](../model_api#intermediate-features) | ✓ |
| [Any input size, no changes needed](../model_api#input-size) | ✓ |
| [`set_input_size()`](../model_api#input-size) | – |
| [`dynamic_img_size=True`](../model_api#input-size) | – |
| [`set_grad_checkpointing()`](../model_api#training-hooks) | ✓ |
| [Layer-wise LR decay (`group_matcher()`)](../model_api#training-hooks) | ✓ |
| [`reset_classifier()`](../model_api#classifier) | ✓ |

✓ supported, ✓ (square) works for square sizes only, ✗ not supported, – not needed (the model already accepts any input
size). Checked by running each feature on a randomly initialized model (`MobileNetV3` with `hardcorenas_b` at 224),
other architectures of the same class are expected to behave the same. See the [Model API](../model_api) page for
details.

## Architectures

| Architecture | Params (M) | GMACs | Input size | Pretrained weights |
|---|---:|---:|---:|---:|
| `hardcorenas_a` | 5.26 | 0.23 | 224 | 1 |
| `hardcorenas_b` | 5.18 | 0.26 | 224 | 1 |
| `hardcorenas_c` | 5.52 | 0.28 | 224 | 1 |
| `hardcorenas_d` | 7.50 | 0.30 | 224 | 1 |
| `hardcorenas_e` | 8.07 | 0.35 | 224 | 1 |
| `hardcorenas_f` | 8.20 | 0.35 | 224 | 1 |

GMACs and parameter counts are for the default input size and classifier of each architecture, from the [benchmark
results](https://github.com/huggingface/pytorch-image-models/tree/main/results).

## Pretrained weights

Top-1 is ImageNet-1k validation accuracy from
[`results-imagenet.csv`](https://github.com/huggingface/pytorch-image-models/blob/main/results/results-imagenet.csv) at
the listed eval size, - if the weight has no ImageNet-1k result (e.g. pretraining or image-text weights). Classes is the
size of the pretrained classifier, none for weights without one.

| Weight | Input size | Top-1 (%) | Eval size | Classes | License | Hub |
|---|---:|---:|---:|---:|---|---|
| `hardcorenas_a.miil_green_in1k` | 224 | 75.926 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/hardcorenas_a.miil_green_in1k) |
| `hardcorenas_b.miil_green_in1k` | 224 | 76.550 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/hardcorenas_b.miil_green_in1k) |
| `hardcorenas_c.miil_green_in1k` | 224 | 77.092 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/hardcorenas_c.miil_green_in1k) |
| `hardcorenas_d.miil_green_in1k` | 224 | 77.450 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/hardcorenas_d.miil_green_in1k) |
| `hardcorenas_e.miil_green_in1k` | 224 | 77.788 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/hardcorenas_e.miil_green_in1k) |
| `hardcorenas_f.miil_green_in1k` | 224 | 78.100 | 224 | 1000 | apache-2.0 | [timm](https://huggingface.co/timm/hardcorenas_f.miil_green_in1k) |