Upload PANPE XRR reflectometry model weights and configuration

LICENSE.txt

@@ -0,0 +1,7 @@
+Copyright (c) 2024 Vladimir Starostin
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

README.md

@@ -1,3 +1,29 @@
----
-license: mit
----
+# PANPE Reflectometry Model Weights
+
+Pre-trained model weights for **Fast and Reliable Probabilistic Reflectometry Inversion with Prior-Amortized Neural Posterior Estimation**.
+
+**Note**: This repository provides model weights for community access. The original work and model training were performed by Vladimir Starostin and colleagues. This is not an official repository by the original authors, but rather a community contribution to make the pre-trained weights easily accessible via HuggingFace.
+
+## Model Description
+
+This repository contains the trained neural network weights for the PANPE (Prior-Amortized Neural Posterior Estimation) model designed for Bayesian reflectometry analysis. The model enables fast and reliable probabilistic inversion of X-ray reflectometry data.
+
+
+### Using the Model
+
+To use these weights, you need the full PANPE package. Get the complete code and installation instructions from:
+
+- **GitHub Repository**: https://github.com/mlcolab/panpe
+- **Zenodo Dataset**: https://zenodo.org/records/14267737
+
+
+## Files in this Repository
+
+- `saved_models/model_panpe-2layers-xrr.pt`: Pre-trained PyTorch model weights
+- `configs/panpe-2layers-xrr.yaml`: Model configuration file
+- `LICENSE.txt`: MIT License
+
+
+## License
+
+This project is licensed under the MIT License - see the [LICENSE.txt](LICENSE.txt) file for details.

configs/panpe-2layers-xrr.yaml

@@ -0,0 +1,75 @@
+general:
+  name: panpe-2layers-xrr
+  root_dir: null
+  device: cuda
+simulator:
+  num_layers: 2
+  hyperprior:
+    cls: HyperpriorForUniformPriorsWithConstrainedRoughness
+    kwargs:
+      thickness_range: [0., 500.]
+      roughness_range: [0., 50.]
+      sld_range: [0., 60.]
+      q_offset_range: [-0.002, 0.002]
+      r_scale_range: [0.95, 1.05]
+      coef: 0.5
+  q_simulator:
+    cls: RandomQSimulator
+    kwargs:
+      q_range: [0.001, 0.15]
+      max_q_num: 64
+      drop_range: [0.0, 0.7]
+  measurement_noise:
+    cls: NormalNoiseSimulator
+    kwargs:
+      rel_sigma_range: [0.1, 0.3]
+  physical_model:
+    cls: BasicLayerStructureModel
+    kwargs:
+      enable_q_misalignment: true
+      enable_r_misalignment: true
+      enable_background: false
+nn:
+  embedding_net:
+    cls: EmbeddingNN
+    pretrained_name: null
+    kwargs:
+       z_num: [16, 32, 64 ]
+       z_range: [ 0., 0.16 ]
+       latent_dim: 256
+       num_blocks: 4
+       kernel_coef: 16
+       use_layer_norm: true
+  flow:
+    num_layers: 40
+    use_batch_norm_transform: true
+    transform_net:
+      context_features: 256
+      hidden_features: 256
+      num_blocks: 3
+      use_layer_norm: true
+training:
+  num_iterations: 300000
+  batch_size: 8192
+  lr: 1.0e-5
+  update_tqdm_freq: 1
+  grad_accumulation_steps: 1
+  optimizer: AdamW
+
+  callbacks:
+    save_best_model:
+      enable: true
+      freq: 500
+    lr_scheduler:
+      cls: StepLR
+      kwargs:
+          step_size: 50000
+          gamma: 0.5
+          last_epoch: 200000
+    lr_warmup:
+      cls: WarmupLR
+      kwargs:
+        start_lr: 1.0e-5
+        end_lr: 2.0e-4
+        warmup_steps: 2000
+        logscale: true

saved_models/model_panpe-2layers-xrr.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:96c0c241556d9df9626d4e35ded607e4a5756c7d99fdc75b2f2919ad5545ab6d
+size 129186122