polymathic-ai/UNetConvNext-active_matter

1---2datasets: polymathic-ai/active_matter3tags:4- physics5---6 7# Benchmarking Models on the Well8 9[The Well](https://github.com/PolymathicAI/the_well) is a 15TB dataset collection of physics simulations. This model is part of the models that have been benchmarked on the Well.10 11 12The models have been trained for a fixed time of 12 hours or up to 500 epochs, whichever happens first. The training was performed on a NVIDIA H100 96GB GPU.13In the time dimension, the context length was set to 4. The batch size was set to maximize the memory usage. We experiment with 5 different learning rates for each model on each dataset.14We use the model performing best on the validation set to report test set results.15 16The reported results are here to provide a simple baseline. **They should not be considered as state-of-the-art**. We hope that the community will build upon these results to develop better architectures for PDE surrogate modeling.17 18# CNextU-Net19 20Implementation of the [U-Net model](https://arxiv.org/abs/1505.04597) using [ConvNext blocks](https://arxiv.org/abs/2201.03545).21 22## Model Details23 24For benchmarking on the Well, we used the following parameters.25 26| Parameters          | Values |27|---------------------|--------|28| Spatial Filter Size | 7      |29| Initial Dimension   | 42     |30| Block per Stage     | 2      |31| Up/Down Blocks      | 4      |32| Bottleneck Blocks   | 1      |33 34 35## Trained Model Versions36 37Below is the list of checkpoints available for the training of CNextU-Net on different datasets of the Well.38 39| Dataset | Learning Rate | Epoch | VRMSE |40|---------|---------------|-------|-------|41| [acoustic_scattering_maze](https://huggingface.co/polymathic-ai/UNetConvNext-acoustic_scattering) | 1E-3 | 10 | 0.0196 |42| [active_matter](https://huggingface.co/polymathic-ai/UNetConvNext-active_matter) | 5E-3 | 156 | 0.0953 |43| [convective_envelope_rsg](https://huggingface.co/polymathic-ai/UNetConvNext-convective_envelope_rsg) | 1E-4 | 5 | 0.0663 |44| [gray_scott_reaction_diffusion](https://huggingface.co/polymathic-ai/UNetConvNext-gray_scott_reaction_diffusion) | 1E-4 | 15 | 0.3596 |45| [helmholtz_staircase](https://huggingface.co/polymathic-ai/UNetConvNext-helmholtz_staircase) | 5E-4 | 47 | 0.00146 |46| [MHD_64](https://huggingface.co/polymathic-ai/UNetConvNext-MHD_64) | 5E-3 | 59 | 0.1487 |47| [planetswe](https://huggingface.co/polymathic-ai/UNetConvNext-planetswe) | 1E-2 | 18 | 0.3268 |48| [post_neutron_star_merger](https://huggingface.co/polymathic-ai/UNetConvNext-post_neutron_star_merger) | - | - | - |49| [rayleigh_benard](https://huggingface.co/polymathic-ai/UNetConvNext-rayleigh_benard) | 5E-4 | 12 | 0.4807 |50| [rayleigh_taylor_instability](https://huggingface.co/polymathic-ai/UNetConvNext-rayleigh_taylor_instability) | 5E-3 | 56 | 0.3771 |51| [shear_flow](https://huggingface.co/polymathic-ai/UNetConvNext-shear_flow) | 5E-4 | 9 | 0.3972 |52| [supernova_explosion_64](https://huggingface.co/polymathic-ai/UNetConvNext-supernova_explosion_64) | 5E-4 | 13 | 0.2801 |53| [turbulence_gravity_cooling](https://huggingface.co/polymathic-ai/UNetConvNext-turbulence_gravity_cooling) | 1E-3 | 3 | 0.2093 |54| [turbulent_radiative_layer_2D](https://huggingface.co/polymathic-ai/UNetConvNext-turbulent_radiative_layer_2D) | 5E-3 | 495 | 0.1247 |55| [viscoelastic_instability](https://huggingface.co/polymathic-ai/UNetConvNext-viscoelastic_instability) | 5E-4 | 114 | 0.1966 |56 57 58## Loading the model from Hugging Face59 60To load the UNetConvNext model trained on the `active_matter` of the Well, use the following commands.61 62```python63from the_well.benchmark.models import UNetConvNext64 65model = UNetConvNext.from_pretrained("polymathic-ai/UNetConvNext-active_matter")66```