updates

Author: Programmer-RD-AI

ML/helper_functions/train.py

@@ -14,7 +14,7 @@ def __init__(
         valid_dataloader: DataLoader,
         criterion: torch.nn,
         optimizer: torch.optim,
-        lr_schedular: bool = None,
+        lr_schedular=None,
     ) -> None:
         self.model = model
         self.epochs = epochs
@@ -75,4 +75,9 @@ def train(self, run_name: str) -> None:
         self.save_model(run_name)
 
     def save_model(self, run_name: str) -> None:
-        pass
+        if run_name not in os.listdir("./ML/predictions/"):
+            os.mkdir(f"./ML/predictions/{run_name}")
+        torch.save(self.model, f"./ML/predictions/{run_name}/model.pt")
+        torch.save(self.model, f"./ML/predictions/{run_name}/model.pth")
+        torch.save(self.model.state_dict(), f"./ML/predictions/{run_name}/model_state_dict.pt")
+        torch.save(self.model.state_dict(), f"./ML/predictions/{run_name}/model_state_dict.pth")

ML/helper_functions/transformations/transformer.py

@@ -38,5 +38,5 @@ def transform(self) -> torchtext.transforms.Sequential:
         )
         return t
 
-    def model_transform(self, model) -> torchtext.transforms:
-        return model.transforms()
+    def model_transform(self, model=XLMR_BASE_ENCODER) -> torchtext.transforms:
+        return model.transform()

run.py

@@ -1,24 +1,33 @@
 from ML import *
 
-lrs = [1e-0, 1e-1, 1e-2, 1e-3, 1e-4, 1e-5, 1e-6]
-for lr in lrs:
+
+def train(
+    batch_size: int = 32,
+    lr: float = 0.01,
+    test_split: float = 0.25,
+    optimizer=optim.Adam,
+    epochs: int = 5,
+    name: str = "",
+    lr_schedular=None,
+    transforms=None,
+):
     train_data_loader, test_data_loader, valid_data_loader = Load_Data(
         Main_DL,
         Valid_Loader,
         [
             "/media/user/Main/Programmer-RD-AI/Programming/Learning/JS/NLP-Disaster-Tweets/ML/data/train.csv",
-            32,
-            Transformer().transform(),
+            batch_size,
+            transforms,
         ],
         [
             "/media/user/Main/Programmer-RD-AI/Programming/Learning/JS/NLP-Disaster-Tweets/ML/data/test.csv",
             1,
         ],
-        0.25,
+        test_split,
         42,
     ).ld()
     model = TL().to(device)
-    optimizer = optim.Adam(model.parameters(), lr=lr)
+    optimizer = optimizer(model.parameters(), lr=lr)
     criterion = nn.CrossEntropyLoss()
     config = {
         "model": model,
@@ -28,11 +37,58 @@
     }
     Train(
         model,
-        5,
+        epochs,
         config,
         train_data_loader,
         test_data_loader,
         valid_data_loader,
         criterion,
         optimizer,
-    ).train(f"{lr}")
+    ).train(f"{name}")
+
+
+train(
+    transforms=Transformer().transform(),
+    batch_size=16,
+    lr=1e-3,
+    test_split=0.25,
+    optimizer=optim.Adam,
+    lr_schedular=None,
+    name=f"1e-3",
+)
+train(
+    transforms=Transformer().transform(),
+    batch_size=16,
+    lr=1e-4,
+    test_split=0.25,
+    optimizer=optim.Adam,
+    lr_schedular=None,
+    name=f"1e-4",
+)
+train(
+    transforms=Transformer().transform(),
+    batch_size=16,
+    lr=1e-5,
+    test_split=0.25,
+    optimizer=optim.Adam,
+    lr_schedular=None,
+    name=f"1e-5",
+)
+train(
+    transforms=Transformer().transform(),
+    batch_size=16,
+    lr=1e-6,
+    test_split=0.25,
+    optimizer=optim.Adam,
+    lr_schedular=None,
+    name=f"1e-6",
+)
+train(
+    transforms=Transformer().transform(),
+    batch_size=16,
+    lr=1e-7,
+    test_split=0.25,
+    optimizer=optim.Adam,
+    lr_schedular=None,
+    name=f"1e-7",
+)

wandb/latest-run

@@ -1 +1 @@
-run-20230727_223900-960s76vj
+run-20230728_100733-nffptso0