add experiments

experiments/config.yaml

@@ -1,5 +1,5 @@
 # lightning.pytorch==2.5.5
-seed_everything: true
+seed_everything: 1999
 trainer:
   accelerator: gpu
   strategy: auto
@@ -14,7 +14,13 @@ trainer:
         mode: min
         save_top_k: 1
         filename: best-checkpoint
-  max_epochs: 50
+    - class_path: lightning.pytorch.callbacks.EarlyStopping
+      init_args:
+        monitor: val/loss
+        mode: min
+        patience: 10
+        verbose: false
+  max_epochs: 200
   min_epochs: null
   max_steps: -1
   min_steps: null
@@ -22,6 +28,8 @@ trainer:
   log_every_n_steps: null
   inference_mode: true
   default_root_dir: null
+  accumulate_grad_batches: 2
+  gradient_clip_val: 1.0
 model: 
   class_path: ThermalSolver.module.GraphSolver
   init_args:
@@ -30,16 +38,16 @@ model:
       x_ch_node: 1 
       f_ch_node: 1 
       hidden: 16
-      layers: 2
+      layers: 1
       edge_ch: 3 
       out_ch: 1
-    unrolling_steps: 10
+    unrolling_steps: 64
 data: 
   class_path: ThermalSolver.data_module.GraphDataModule
   init_args:
     hf_repo: "SISSAmathLab/thermal-conduction"
-    split_name: "2000"
+    split_name: "2000_ref_1"
-    batch_size: 6
+    batch_size: 10
     train_size: 0.8
     test_size: 0.1
     test_size: 0.1

experiments/config_01.yaml

@@ -0,0 +1,61 @@
+# lightning.pytorch==2.5.5
+seed_everything: 1999
+trainer:
+  accelerator: gpu
+  strategy: auto
+  devices: 1
+  num_nodes: 1
+  precision: null
+  logger:
+    - class_path: lightning.pytorch.loggers.TensorBoardLogger
+      init_args:
+        save_dir: lightning_logs
+        name: "01"
+        version: null
+  callbacks:
+    - class_path: lightning.pytorch.callbacks.ModelCheckpoint
+      init_args:
+        monitor: val/loss
+        mode: min
+        save_top_k: 1
+        filename: best-checkpoint
+    - class_path: lightning.pytorch.callbacks.EarlyStopping
+      init_args:
+        monitor: val/loss
+        mode: min
+        patience: 15
+        verbose: false
+  max_epochs: 1000
+  min_epochs: null
+  max_steps: -1
+  min_steps: null
+  overfit_batches: 0.0
+  log_every_n_steps: null
+  inference_mode: true
+  default_root_dir: null
+  accumulate_grad_batches: 6
+  # gradient_clip_val: 1.0
+model: 
+  class_path: ThermalSolver.module.GraphSolver
+  init_args:
+    model_class_path: ThermalSolver.model.local_gno.GatingGNO
+    model_init_args:
+      x_ch_node: 1 
+      f_ch_node: 1 
+      hidden: 16
+      layers: 1
+      edge_ch: 3 
+      out_ch: 1
+    unrolling_steps: 64
+data: 
+  class_path: ThermalSolver.data_module.GraphDataModule
+  init_args:
+    hf_repo: "SISSAmathLab/thermal-conduction"
+    split_name: "2000_ref_1"
+    batch_size: 4
+    train_size: 0.8
+    test_size: 0.1
+    test_size: 0.1
+optimizer: null
+lr_scheduler: null
+# ckpt_path: lightning_logs/01/version_0/checkpoints/best-checkpoint.ckpt

experiments/config_02.yaml

@@ -0,0 +1,62 @@
+# lightning.pytorch==2.5.5
+seed_everything: 1999
+trainer:
+  accelerator: gpu
+  strategy: auto
+  devices: 1
+  num_nodes: 1
+  precision: null
+  logger:
+    - class_path: lightning.pytorch.loggers.TensorBoardLogger
+      init_args:
+        save_dir: lightning_logs
+        name: "02"
+        version: null
+  callbacks:
+    - class_path: lightning.pytorch.callbacks.ModelCheckpoint
+      init_args:
+        monitor: val/loss
+        mode: min
+        save_top_k: 1
+        filename: best-checkpoint
+    - class_path: lightning.pytorch.callbacks.EarlyStopping
+      init_args:
+        monitor: val/loss
+        mode: min
+        patience: 10
+        verbose: false
+  max_epochs: 200
+  min_epochs: null
+  max_steps: -1
+  min_steps: null
+  overfit_batches: 0.0
+  log_every_n_steps: null
+  inference_mode: true
+  default_root_dir: null
+  accumulate_grad_batches: 6
+  gradient_clip_val: 1.0
+model: 
+  class_path: ThermalSolver.module.GraphSolver
+  init_args:
+    model_class_path: ThermalSolver.model.local_gno.GatingGNO
+    model_init_args:
+      x_ch_node: 1 
+      f_ch_node: 1 
+      hidden: 16
+      layers: 2
+      edge_ch: 3 
+      out_ch: 1
+    unrolling_steps: 32
+data: 
+  class_path: ThermalSolver.data_module.GraphDataModule
+  init_args:
+    hf_repo: "SISSAmathLab/thermal-conduction"
+    split_name: "2000_ref_1"
+    batch_size: 4
+    train_size: 0.8
+    test_size: 0.1
+    test_size: 0.1
+optimizer: null
+lr_scheduler: null
+# ckpt_path: lightning_logs/version_15/checkpoints/best-checkpoint.ckpt
+ckpt_path: null

experiments/config_04.yaml

@@ -0,0 +1,62 @@
+# lightning.pytorch==2.5.5
+seed_everything: 1999
+trainer:
+  accelerator: gpu
+  strategy: auto
+  devices: 1
+  num_nodes: 1
+  precision: null
+  logger:
+    - class_path: lightning.pytorch.loggers.TensorBoardLogger
+      init_args:
+        save_dir: lightning_logs
+        name: "04"
+        version: null
+  callbacks:
+    - class_path: lightning.pytorch.callbacks.ModelCheckpoint
+      init_args:
+        monitor: val/loss
+        mode: min
+        save_top_k: 1
+        filename: best-checkpoint
+    - class_path: lightning.pytorch.callbacks.EarlyStopping
+      init_args:
+        monitor: val/loss
+        mode: min
+        patience: 10
+        verbose: false
+  max_epochs: 200
+  min_epochs: null
+  max_steps: -1
+  min_steps: null
+  overfit_batches: 0.0
+  log_every_n_steps: null
+  inference_mode: true
+  default_root_dir: null
+  accumulate_grad_batches: 6
+  gradient_clip_val: 1.0
+model: 
+  class_path: ThermalSolver.module.GraphSolver
+  init_args:
+    model_class_path: ThermalSolver.model.local_gno.GatingGNO
+    model_init_args:
+      x_ch_node: 1 
+      f_ch_node: 1 
+      hidden: 16
+      layers: 4
+      edge_ch: 3 
+      out_ch: 1
+    unrolling_steps: 16
+data: 
+  class_path: ThermalSolver.data_module.GraphDataModule
+  init_args:
+    hf_repo: "SISSAmathLab/thermal-conduction"
+    split_name: "2000_ref_1"
+    batch_size: 4
+    train_size: 0.8
+    test_size: 0.1
+    test_size: 0.1
+optimizer: null
+lr_scheduler: null
+# ckpt_path: lightning_logs/version_15/checkpoints/best-checkpoint.ckpt
+ckpt_path: null

experiments/config_08.yaml

@@ -0,0 +1,62 @@
+# lightning.pytorch==2.5.5
+seed_everything: 1999
+trainer:
+  accelerator: gpu
+  strategy: auto
+  devices: 1
+  num_nodes: 1
+  precision: null
+  logger:
+    - class_path: lightning.pytorch.loggers.TensorBoardLogger
+      init_args:
+        save_dir: lightning_logs
+        name: "08"
+        version: null
+  callbacks:
+    - class_path: lightning.pytorch.callbacks.ModelCheckpoint
+      init_args:
+        monitor: val/loss
+        mode: min
+        save_top_k: 1
+        filename: best-checkpoint
+    - class_path: lightning.pytorch.callbacks.EarlyStopping
+      init_args:
+        monitor: val/loss
+        mode: min
+        patience: 10
+        verbose: false
+  max_epochs: 200
+  min_epochs: null
+  max_steps: -1
+  min_steps: null
+  overfit_batches: 0.0
+  log_every_n_steps: null
+  inference_mode: true
+  default_root_dir: null
+  accumulate_grad_batches: 6
+  gradient_clip_val: 1.0
+model: 
+  class_path: ThermalSolver.module.GraphSolver
+  init_args:
+    model_class_path: ThermalSolver.model.local_gno.GatingGNO
+    model_init_args:
+      x_ch_node: 1 
+      f_ch_node: 1 
+      hidden: 16
+      layers: 8
+      edge_ch: 3 
+      out_ch: 1
+    unrolling_steps: 8
+data: 
+  class_path: ThermalSolver.data_module.GraphDataModule
+  init_args:
+    hf_repo: "SISSAmathLab/thermal-conduction"
+    split_name: "2000_ref_1"
+    batch_size: 4
+    train_size: 0.8
+    test_size: 0.1
+    test_size: 0.1
+optimizer: null
+lr_scheduler: null
+# ckpt_path: lightning_logs/version_15/checkpoints/best-checkpoint.ckpt
+ckpt_path: null

experiments/config_16.yaml

@@ -0,0 +1,62 @@
+# lightning.pytorch==2.5.5
+seed_everything: 1999
+trainer:
+  accelerator: gpu
+  strategy: auto
+  devices: 1
+  num_nodes: 1
+  precision: null
+  logger:
+    - class_path: lightning.pytorch.loggers.TensorBoardLogger
+      init_args:
+        save_dir: lightning_logs
+        name: "16"
+        version: null
+  callbacks:
+    - class_path: lightning.pytorch.callbacks.ModelCheckpoint
+      init_args:
+        monitor: val/loss
+        mode: min
+        save_top_k: 1
+        filename: best-checkpoint
+    - class_path: lightning.pytorch.callbacks.EarlyStopping
+      init_args:
+        monitor: val/loss
+        mode: min
+        patience: 10
+        verbose: false
+  max_epochs: 200
+  min_epochs: null
+  max_steps: -1
+  min_steps: null
+  overfit_batches: 0.0
+  log_every_n_steps: null
+  inference_mode: true
+  default_root_dir: null
+  accumulate_grad_batches: 6
+  gradient_clip_val: 1.0
+model: 
+  class_path: ThermalSolver.module.GraphSolver
+  init_args:
+    model_class_path: ThermalSolver.model.local_gno.GatingGNO
+    model_init_args:
+      x_ch_node: 1 
+      f_ch_node: 1 
+      hidden: 16
+      layers: 16
+      edge_ch: 3 
+      out_ch: 1
+    unrolling_steps: 4
+data: 
+  class_path: ThermalSolver.data_module.GraphDataModule
+  init_args:
+    hf_repo: "SISSAmathLab/thermal-conduction"
+    split_name: "2000_ref_1"
+    batch_size: 4
+    train_size: 0.8
+    test_size: 0.1
+    test_size: 0.1
+optimizer: null
+lr_scheduler: null
+# ckpt_path: lightning_logs/version_15/checkpoints/best-checkpoint.ckpt
+ckpt_path: null

experiments/config_32.yaml

@@ -0,0 +1,62 @@
+# lightning.pytorch==2.5.5
+seed_everything: 1999
+trainer:
+  accelerator: gpu
+  strategy: auto
+  devices: 1
+  num_nodes: 1
+  precision: null
+  logger:
+    - class_path: lightning.pytorch.loggers.TensorBoardLogger
+      init_args:
+        save_dir: lightning_logs
+        name: "32"
+        version: null
+  callbacks:
+    - class_path: lightning.pytorch.callbacks.ModelCheckpoint
+      init_args:
+        monitor: val/loss
+        mode: min
+        save_top_k: 1
+        filename: best-checkpoint
+    - class_path: lightning.pytorch.callbacks.EarlyStopping
+      init_args:
+        monitor: val/loss
+        mode: min
+        patience: 10
+        verbose: false
+  max_epochs: 200
+  min_epochs: null
+  max_steps: -1
+  min_steps: null
+  overfit_batches: 0.0
+  log_every_n_steps: null
+  inference_mode: true
+  default_root_dir: null
+  accumulate_grad_batches: 6
+  gradient_clip_val: 1.0
+model: 
+  class_path: ThermalSolver.module.GraphSolver
+  init_args:
+    model_class_path: ThermalSolver.model.local_gno.GatingGNO
+    model_init_args:
+      x_ch_node: 1 
+      f_ch_node: 1 
+      hidden: 16
+      layers: 32
+      edge_ch: 3 
+      out_ch: 1
+    unrolling_steps: 2
+data: 
+  class_path: ThermalSolver.data_module.GraphDataModule
+  init_args:
+    hf_repo: "SISSAmathLab/thermal-conduction"
+    split_name: "2000_ref_1"
+    batch_size: 4
+    train_size: 0.8
+    test_size: 0.1
+    test_size: 0.1
+optimizer: null
+lr_scheduler: null
+# ckpt_path: lightning_logs/version_15/checkpoints/best-checkpoint.ckpt
+ckpt_path: null

experiments/config_64.yaml

@@ -0,0 +1,63 @@
+# lightning.pytorch==2.5.5
+seed_everything: 1999
+trainer:
+  accelerator: gpu
+  strategy: auto
+  devices: 1
+  num_nodes: 1
+  precision: null
+  logger:
+    - class_path: lightning.pytorch.loggers.TensorBoardLogger
+      init_args:
+        save_dir: lightning_logs
+        name: "64"
+        version: null
+  callbacks:
+    - class_path: lightning.pytorch.callbacks.ModelCheckpoint
+      init_args:
+        monitor: val/loss
+        mode: min
+        save_top_k: 1
+        filename: best-checkpoint
+    - class_path: lightning.pytorch.callbacks.EarlyStopping
+      init_args:
+        monitor: val/loss
+        mode: min
+        patience: 15
+        verbose: false
+  max_epochs: 1000
+  min_epochs: null
+  max_steps: -1
+  min_steps: null
+  overfit_batches: 0.0
+  log_every_n_steps: null
+  inference_mode: true
+  default_root_dir: null
+  accumulate_grad_batches: 6
+  gradient_clip_val: 1.0
+
+model: 
+  class_path: ThermalSolver.module.GraphSolver
+  init_args:
+    model_class_path: ThermalSolver.model.local_gno.GatingGNO
+    model_init_args:
+      x_ch_node: 1 
+      f_ch_node: 1 
+      hidden: 16
+      layers: 64
+      edge_ch: 3 
+      out_ch: 1
+    unrolling_steps: 1
+data: 
+  class_path: ThermalSolver.data_module.GraphDataModule
+  init_args:
+    hf_repo: "SISSAmathLab/thermal-conduction"
+    split_name: "2000"
+    batch_size: 4
+    train_size: 0.8
+    test_size: 0.1
+    test_size: 0.1
+optimizer: null
+lr_scheduler: null
+# ckpt_path: lightning_logs/64/version_0/checkpoints/best-checkpoint.ckpt
+ckpt_path: null