Hotfix/sg 645 regression tests essential fixes (#669)

* release tag removed for check

* missing ckpt root in recipes fixed

* warn instead of error if max batches longer then loader

* .__version__update

* release tag filter added back

src/super_gradients/__init__.py

@@ -22,6 +22,6 @@
     "setup_device",
 ]
 
-__version__ = "3.0.6"
+__version__ = "3.0.7"
 
 env_sanity_check()

src/super_gradients/recipes/imagenet_mobilenetv3_large.yaml

@@ -20,7 +20,7 @@ arch_params:
 experiment_name: mobileNetv3_large_training
 
 architecture: mobilenet_v3_large
-
+ckpt_root_dir:
 
 # THE FOLLOWING PARAMS ARE DIRECTLY USED BY HYDRA
 hydra:

src/super_gradients/recipes/imagenet_mobilenetv3_small.yaml

@@ -20,7 +20,7 @@ arch_params:
 experiment_name: mobileNetv3_small_training
 
 architecture: mobilenet_v3_small
-
+ckpt_root_dir:
 
 # THE FOLLOWING PARAMS ARE DIRECTLY USED BY HYDRA
 hydra:

src/super_gradients/recipes/imagenet_repvgg.yaml

@@ -37,7 +37,7 @@ multi_gpu: DDP
 num_gpus: 4
 
 architecture: repvgg_a0
-
+ckpt_root_dir:
 
 # THE FOLLOWING PARAMS ARE DIRECTLY USED BY HYDRA
 hydra:

src/super_gradients/recipes/imagenet_vit_base.yaml

@@ -33,7 +33,7 @@ experiment_name: vit_base_imagenet1k
 architecture: vit_base
 multi_gpu: DDP
 num_gpus: 8
-
+ckpt_root_dir:
 
 # THE FOLLOWING PARAMS ARE DIRECTLY USED BY HYDRA
 hydra:

src/super_gradients/recipes/imagenet_vit_large.yaml

@@ -29,7 +29,7 @@ architecture: vit_large
 experiment_name: vit_large_imagenet1k
 multi_gpu: DDP
 num_gpus: 8
-
+ckpt_root_dir:
 
 # THE FOLLOWING PARAMS ARE DIRECTLY USED BY HYDRA
 hydra:

src/super_gradients/training/sg_trainer/sg_trainer.py

@@ -1192,19 +1192,19 @@ def forward(self, inputs, targets):
 
         self.ckpt_best_name = self.training_params.ckpt_best_name
 
-        if self.training_params.max_train_batches is not None and (
-            self.training_params.max_train_batches > len(self.train_loader) or self.training_params.max_train_batches <= 0
-        ):
-
-            raise ValueError("max_train_batches must be positive and smaller then len(train_loader).")
+        if self.training_params.max_train_batches is not None:
+            if self.training_params.max_train_batches > len(self.train_loader):
+                logger.warning("max_train_batches is greater than len(self.train_loader) and will have no effect.")
+            elif self.training_params.max_train_batches <= 0:
+                raise ValueError("max_train_batches must be positive.")
+
+        if self.training_params.max_valid_batches is not None:
+            if self.training_params.max_valid_batches > len(self.valid_loader):
+                logger.warning("max_valid_batches is greater than len(self.valid_loader) and will have no effect.")
+            elif self.training_params.max_valid_batches <= 0:
+                raise ValueError("max_valid_batches must be positive.")
 
         self.max_train_batches = self.training_params.max_train_batches
-
-        if self.training_params.max_valid_batches is not None and (
-            self.training_params.max_valid_batches > len(self.valid_loader) or self.training_params.max_valid_batches <= 0
-        ):
-
-            raise ValueError("max_valid_batches must be positive and smaller then len(valid_loader).")
         self.max_valid_batches = self.training_params.max_valid_batches
 
         # STATE ATTRIBUTE SET HERE FOR SUBSEQUENT TRAIN() CALLS