Upload G-PARC model weights, test data, and configs (4 models)

README.md

@@ -11,21 +11,14 @@ tags:
 
 Model weights, test data, and configuration files for the G-PARC elastoplastic simulation paper.
 
-## Contents
+## Models
 
-```
-checkpoints/          # Trained model weights (.pth)
-  gparcv1_best.pth    # G-PARCv1 (GNN baseline)
-  gparcv2_best.pth    # G-PARCv2 (MLS + numerical integration) 
-  meshgraphkan_best.pth  # MeshGraphKAN (NVIDIA architecture)
-
-configs/              # Training configuration files
-data/
-  normalization_stats.json   # Global max normalization parameters
-  test/                      # PLAID test simulations (.pt)
-
-training_histories/   # Loss curves per epoch
-```
+| Model | Description |
+|-------|-------------|
+| G-PARCv1 | Graph Physics-Aware Recurrent Convolutions — fully learned GNN operators |
+| G-PARCv2 | MLS differential operators + numerical Euler integration |
+| MeshGraphKAN | Kolmogorov-Arnold Network message passing with Fourier basis |
+| MeshGraphNet | Standard encode-process-decode GNN (Pfaff et al., 2021) |
 
 ## Dataset
 
@@ -39,10 +32,5 @@ PLAID 2D Elasto-Plasto-Dynamics benchmark — high-velocity impact on steel plat
 
 ```python
 from huggingface_hub import hf_hub_download
-
 ckpt = hf_hub_download("jacktbeerman/Gparc", "checkpoints/gparcv2_best.pth")
 ```
-
-## Code
-
-Full training and evaluation code: [GitHub repo link here]

checkpoints/meshgraphkan_best.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2910ac12140eef16eb0f3ab9c5946e132ed197eac1ce7c5cacf93f663475536e
+oid sha256:e4d3fa10ae1104788402b4bb275ec2ff670e3c9c2722e22b52f88c70709f3e5f
 size 8109521

checkpoints/meshgraphnet_best.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1cfcb1561e5c8bfc247069608156a27cf14589b7fe9221da17983274915be8ba
+size 6250193

configs/meshgraphkan_config.json

@@ -26,7 +26,7 @@
   "grad_clip_norm": 1.0,
   "device": "cuda",
   "num_workers": 4,
-  "output_dir": "/scratch/jtb3sud/elasto_meshgraphkan/run1",
+  "output_dir": "/scratch/jtb3sud/delta/elasto",
   "resume": null,
   "reset_best": false,
   "fresh_scheduler": false,

configs/meshgraphnet_config.json

@@ -0,0 +1,21 @@
+{
+  "data_dir": "/scratch/jtb3sud/processed_elasto_plastic/global_max/normalized/small",
+  "seq_len": 16,
+  "stride": 16,
+  "num_static_feats": 2,
+  "num_dynamic_feats": 2,
+  "hidden_dim": 128,
+  "num_layers": 4,
+  "epochs": 1500,
+  "lr": 0.0001,
+  "weight_decay": 0.0005,
+  "grad_clip_norm": 1.0,
+  "scheduler": "cosine",
+  "scheduler_step": 100,
+  "scheduler_gamma": 0.9,
+  "device": "cuda",
+  "num_workers": 4,
+  "checkpoint_dir": "/scratch/jtb3sud/meshgraphnet/elasto/run1",
+  "val_every": 10,
+  "save_every": 100
+}