File size: 7,619 Bytes
d71d2c3 e01daef |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 |
# ViSNet
## Reference
Yusong Wang, Tong Wang, Shaoning Li, Xinheng He, Mingyu Li, Zun Wang, Nanning Zheng, Bin Shao, and Tie-Yan Liu.
Enhancing geometric representations for molecules with equivariant vector-scalar interactive message passing.
Nature Communications, 15(1), January 2024. ISSN: 2041-1723.
URL: https://dx.doi.org/10.1038/s41467-023-43720-2.
## Hyperparameters, model configurations and training strategies
### Model architecture
| Parameter | Value | Description |
|--------------------|----------|--------------------------------------------------------------------------|
| `num_layers` | `4` | Number of ViSNet layers. |
| `num_channels` | `128` | Number of channels. |
| `l_max` | `2` | Highest harmonic order included in the Spherical Harmonics series. |
| `num_heads` | `8` | Number of heads in the attention block. |
| `num_rbf` | `32` | Number of radial basis functions in the embedding block. |
| `trainable_rbf` | `False` | Whether to add learnable weights to the radial embedding basis functions.|
| `activation` | `silu` | Activation function for the output block. |
| `attn_activation` | `silu` | Activation function for the attention block. |
| `vecnorm_type` | `None` | Type of the vector norm. |
| `atomic_energies` | `average`| Treatment of the atomic energies. |
| `avg_um_neighbors` | `None` | Mean number of neighbors. |
### Training
| Parameter | Value | Description |
|--------------------------|--------|--------------------------------------------------|
| `num_epochs` | `220` | Number of epochs to run. |
| `ema_decay` | `0.99` | The EMA decay rate. |
| `eval_num_graphs` | `None` | Number of validation set graphs to evaluate on. |
| `use_ema_params_for_eval`| `True` | Whether to use the EMA parameters for evaluation.|
### Optimizer
| Parameter | Value | Description |
|----------------------------------|----------------|-----------------------------------------------------------------|
| `init_learning_rate` | `0.0001` | Initial learning rate. |
| `peak_learning_rate` | `0.0001` | Peak learning rate. |
| `final_learning_rate` | `0.0001` | Final learning rate. |
| `weight_decay` | `0` | Weight decay. |
| `warmup_steps` | `4000` | Number of optimizer warm-up steps. |
| `transition_steps` | `360000` | Number of optimizer transition steps. |
| `grad_norm` | `500` | Gradient norm used for gradient clipping. |
| `num_gradient_accumulation_steps`| `1` | Steps to accumulate before taking an optimizer step. |
| `algorithm` | `optax.amsgrad`| The AMSGrad optimizer. |
| `b1` | `0.9` | Exponential decay rate to track first moment of past gradients. |
| `b2` | `0.999` | Exponential decay rate to track second moment of past gradients.|
| `eps` | `1e-8` | Constant applied to denominator outside the square root. |
| `eps_root` | `0.0` | Constant applied to denominator inside the square root. |
### Huber Loss Energy weight schedule
| Parameter | Value | Description |
|-----------------------|------------------------------------|-------------------------------------------------------------------------------------------------|
| `schedule` | `optax.piecewise_constant_schedule`| Piecewise constant schedule with scaled jumps at specific boundaries. |
| `init_value` | `40` | Initial value. |
| `boundaries_and_scale`| `{115: 25}` | Dictionary of {step: scale} where scale is multiplied into the schedule value at the given step.|
### Huber Loss Force weight schedule
| Parameter | Value | Description |
|-----------------------|------------------------------------|-------------------------------------------------------------------------------------------------|
| `schedule` | `optax.piecewise_constant_schedule`| Piecewise constant schedule with scaled jumps at specific boundaries. |
| `init_value` | `1000` | Initial value. |
| `boundaries_and_scale`| `{115: 0.04}` | Dictionary of {step: scale} where scale is multiplied into the schedule value at the given step.|
### Dataset
| Parameter | Value | Description |
|-----------------------------|-------|--------------------------------------------|
| `graph_cutoff_angstrom` | `5` | Graph cutoff distance (in Å). |
| `max_n_node` | `32` | Maximum number of nodes allowed in a batch.|
| `max_n_edge` | `288` | Maximum number of edges allowed in a batch.|
| `batch_size` | `16` | Number of graphs in a batch. |
This model was trained on the [SPICE2_curated dataset](https://huggingface.co/datasets/InstaDeepAI/SPICE2-curated).
## How to Use
For complete usage instructions and more information, please refer to our [documentation](https://instadeep.github.io/mlip)
## License summary
1. The Licensed Models are **only** available under this License for Non-Commercial Purposes.
2. You are permitted to reproduce, publish, share and adapt the Output generated by the Licensed Model only for Non-Commercial Purposes and in accordance with this License.
3. You may **not** use the Licensed Models or any of its Outputs in connection with:
1. any Commercial Purposes, unless agreed by Us under a separate licence;
2. to train, improve or otherwise influence the functionality or performance of any other third-party derivative model that is commercial or intended for a Commercial Purpose and is similar to the Licensed Models;
3. to create models distilled or derived from the Outputs of the Licensed Models, unless such models are for Non-Commercial Purposes and open-sourced under the same license as the Licensed Models; or
4. in violation of any applicable laws and regulations.
|