Final debugging steps for pretrained model evaluation

BojarLab · Aug 30, 2024 · 336ad74 · 336ad74
1 parent afcbd83
commit 336ad74
Show file tree

Hide file tree

Showing 3 changed files with 32 additions and 25 deletions.
diff --git a/configs/downstream/pretrained.yaml b/configs/downstream/pretrained.yaml
@@ -25,11 +25,12 @@ datasets:
 #    task: multilabel
 pre-transforms:
   PretrainEmbed:
-    file_path: /scratch/SCRATCH_SAS/roman/Gothenburg/GIFFLAR/data_embed/
+    folder: /scratch/SCRATCH_SAS/roman/Gothenburg/GIFFLAR/data_embed/
     model_name: GIFFLAR
     hash_str: 3fd297ab
 model:
   - name: mlp
+    feat_dim: 1024
     hidden_dim: 1024
     batch_size: 256
     num_layers: 3

diff --git a/gifflar/pretransforms.py b/gifflar/pretransforms.py
@@ -414,15 +414,20 @@ class PretrainEmbed(RootTransform):
     def __init__(self, folder: str, dataset_name: str, model_name: str, hash_str: str, **kwargs: Any):
         super(PretrainEmbed, self).__init__(**kwargs)
         self.data = torch.load(Path(folder) / f"{dataset_name}_{model_name}_{hash_str}.pt")
-        self.lookup = {smiles: (i, j) for i in range(len(self.data)) for j, smiles in enumerate(self.data[i])}
+        self.lookup = {smiles: (i, j) for i in range(len(self.data)) for j, smiles in enumerate(self.data[i]["smiles"])}
         self.pooling = GIFFLARPooling()
         self.layer = kwargs.get("layer", -1)
 
     def __call__(self, data: HeteroData) -> HeteroData:
+        if data["smiles"] not in self.lookup:
+            print(data["smiles"], "not found in preprocessed data.")
+            data["fp"] = torch.zeros_like(data["fp"])
+            return data
+
         a, b = self.lookup[data["smiles"]]
         mask = {key: self.data[a]["batch_ids"][key] == b for key in ["atoms", "bonds", "monosacchs"]}
-        node_embeds = {key: self.data[a]["node_embeds"][self.layer][mask[key]] for key in ["atoms", "bonds", "monosacchs"]}
-        batch_ids = {key: torch.zeros_like(node_embeds[key], dtype=torch.long) for key in ["atoms", "bonds", "monosacchs"]}
+        node_embeds = {key: self.data[a]["node_embeds"][self.layer][key][mask[key]] for key in ["atoms", "bonds", "monosacchs"]}
+        batch_ids = {key: torch.zeros(len(node_embeds[key]), dtype=torch.long) for key in ["atoms", "bonds", "monosacchs"]}
         data["fp"] = self.pooling(node_embeds, batch_ids)
         return data
 

diff --git a/gifflar/train.py b/gifflar/train.py
@@ -232,24 +232,25 @@ def main(config):
 
 
 if __name__ == '__main__':
-    embed(
-        prep_args={
-            "model_name": "GIFFLAR",
-            "ckpt_path": "/scratch/SCRATCH_SAS/roman/Gothenburg/GIFFLAR/logs_pret/gifflar_dyn_re_pretrain/version_0/checkpoints/epoch=99-step=6200.ckpt",
-            "hparams_path": "/scratch/SCRATCH_SAS/roman/Gothenburg/GIFFLAR/logs_pret/gifflar_dyn_re_pretrain/version_0/hparams.yaml",
-            "save_dir":"/scratch/SCRATCH_SAS/roman/Gothenburg/GIFFLAR/data_embed/",
-        },
-        **{
-            "seed": 42,
-            "data_dir": "/scratch/SCRATCH_SAS/roman/Gothenburg/GIFFLAR/",
-            "root_dir": "/scratch/SCRATCH_SAS/roman/Gothenburg/GIFFLAR/data_embed",
-            "logs_dir": "/scratch/SCRATCH_SAS/roman/Gothenburg/GIFFLAR/logs_embed",
-            "dataset": {"name": "Immunogenicity", "task": "classification"},
-            "pre-transforms": {},
-            "hash": "12345678",
-            "model": {},
-        }
-    )
-    # parser = ArgumentParser()
-    # parser.add_argument("config", type=str, help="Path to YAML config file")
-    # main(parser.parse_args().config)
+    #embed(
+    #    prep_args={
+    #        "model_name": "GIFFLAR",
+    #        "ckpt_path": "/scratch/SCRATCH_SAS/roman/Gothenburg/GIFFLAR/logs_pret/gifflar_dyn_re_pretrain/version_0/checkpoints/epoch=99-step=6200.ckpt",
+    #        "hparams_path": "/scratch/SCRATCH_SAS/roman/Gothenburg/GIFFLAR/logs_pret/gifflar_dyn_re_pretrain/version_0/hparams.yaml",
+    #        "save_dir":"/scratch/SCRATCH_SAS/roman/Gothenburg/GIFFLAR/data_embed/",
+    #    },
+    #    **{
+    #        "seed": 42,
+    #        "data_dir": "/scratch/SCRATCH_SAS/roman/Gothenburg/GIFFLAR/",
+    #        "root_dir": "/scratch/SCRATCH_SAS/roman/Gothenburg/GIFFLAR/data_embed",
+    #        "logs_dir": "/scratch/SCRATCH_SAS/roman/Gothenburg/GIFFLAR/logs_embed",
+    #        "dataset": {"name": "Immunogenicity", "task": "classification"},
+    #        "pre-transforms": {},
+    #        "hash": "12345678",
+    #        "model": {},
+    #    }
+    #)
+    parser = ArgumentParser()
+    parser.add_argument("config", type=str, help="Path to YAML config file")
+    main(parser.parse_args().config)
+