Fix --load-checkpoint bug

mlpf/pyg_pipeline.py

@@ -192,6 +192,9 @@ def run(rank, world_size, config, args, outdir, logfile):
             checkpoint_freq=config["checkpoint_freq"],
         )
 
+        checkpoint = torch.load(f"{outdir}/best_weights.pth", map_location=torch.device(rank))
+        model, optimizer = load_checkpoint(checkpoint, model, optimizer)
+
     if args.test:
         if config["load"] is None:
             # if we don't load, we must have a newly trained model
@@ -200,9 +203,6 @@ def run(rank, world_size, config, args, outdir, logfile):
         else:
             outdir = config["load"]
 
-        checkpoint = torch.load(f"{outdir}/best_weights.pth", map_location=torch.device(rank))
-        model, optimizer = load_checkpoint(checkpoint, model, optimizer)
-
         for type_ in config["test_dataset"][config["dataset"]]:  # will be "physical", "gun"
             batch_size = config["test_dataset"][config["dataset"]][type_]["batch_size"] * config["gpu_batch_multiplier"]
             for sample in config["test_dataset"][config["dataset"]][type_]["samples"]:

parameters/pyg-cms-physical.yaml

@@ -0,0 +1,74 @@
+backend: pytorch
+
+dataset: cms
+data_dir:
+gpus: "0"
+gpu_batch_multiplier: 1
+load:
+num_epochs: 2
+patience: 20
+lr: 0.0001
+conv_type: gnn_lsh
+ntrain:
+ntest:
+nvalid: 500
+num_workers: 0
+prefetch_factor:
+checkpoint_freq:
+
+model:
+  gnn_lsh:
+    conv_type: gnn_lsh
+    embedding_dim: 512
+    width: 512
+    num_convs: 3
+    dropout: 0.0
+
+  gravnet:
+    conv_type: gravnet
+    embedding_dim: 512
+    width: 512
+    num_convs: 3
+    k: 16
+    propagate_dimensions: 32
+    space_dimensions: 4
+    dropout: 0.0
+
+  attention:
+    conv_type: attention
+    embedding_dim: 256
+    width: 256
+    num_convs: 3
+    dropout: 0.0
+
+train_dataset:
+  cms:
+    physical:
+      batch_size: 1
+      samples:
+        cms_pf_ttbar:
+          version: 1.6.0
+        cms_pf_qcd:
+          version: 1.6.0
+        cms_pf_ztt:
+          version: 1.6.0
+        cms_pf_qcd_high_pt:
+          version: 1.6.0
+        cms_pf_sms_t1tttt:
+          version: 1.6.0
+
+valid_dataset:
+  cms:
+    physical:
+      batch_size: 1
+      samples:
+        cms_pf_qcd_high_pt:
+          version: 1.6.0
+
+test_dataset:
+  cms:
+    physical:
+      batch_size: 1
+      samples:
+        cms_pf_qcd_high_pt:
+          version: 1.6.0

parameters/pyg-cms-small-highqcd.yaml

@@ -0,0 +1,70 @@
+backend: pytorch
+
+dataset: cms
+data_dir:
+gpus: "0"
+gpu_batch_multiplier: 1
+load:
+num_epochs: 2
+patience: 20
+lr: 0.0001
+conv_type: gnn_lsh
+ntrain:
+ntest:
+nvalid: 500
+num_workers: 0
+prefetch_factor:
+checkpoint_freq:
+
+model:
+  gnn_lsh:
+    conv_type: gnn_lsh
+    embedding_dim: 512
+    width: 512
+    num_convs: 3
+    dropout: 0.0
+
+  gravnet:
+    conv_type: gravnet
+    embedding_dim: 512
+    width: 512
+    num_convs: 3
+    k: 16
+    propagate_dimensions: 32
+    space_dimensions: 4
+    dropout: 0.0
+
+  attention:
+    conv_type: attention
+    embedding_dim: 256
+    width: 256
+    num_convs: 3
+    dropout: 0.0
+
+train_dataset:
+  cms:
+    physical:
+      batch_size: 1
+      samples:
+        cms_pf_ttbar:
+          version: 1.6.0
+        cms_pf_qcd:
+          version: 1.6.0
+        cms_pf_qcd_high_pt:
+          version: 1.6.0
+
+valid_dataset:
+  cms:
+    physical:
+      batch_size: 1
+      samples:
+        cms_pf_qcd_high_pt:
+          version: 1.6.0
+
+test_dataset:
+  cms:
+    physical:
+      batch_size: 1
+      samples:
+        cms_pf_qcd_high_pt:
+          version: 1.6.0