Merge pull request #30 from Genentech/fix-pearson

added reset method to PearsonCorrCoef

Author: avantikalal

src/grelu/lightning/metrics.py

@@ -144,3 +144,6 @@ def compute(self) -> torch.Tensor:
             return output.mean()
         else:
             return output
+
+    def reset(self) -> None:
+        self.pearson.reset()

tests/test_models.py

@@ -12,17 +12,16 @@
     EnformerModel,
     EnformerPretrainedModel,
 )
+from grelu.resources import DEFAULT_WANDB_HOST
 from grelu.sequence.format import convert_input_type
 
 inputs = convert_input_type(["A" * 128], "one_hot")
 
 
 try:
-    wandb.login(host="https://genentech.wandb.io", anonymous="never", timeout=0)
+    wandb.login(host=DEFAULT_WANDB_HOST, anonymous="never", timeout=0)
 except wandb.errors.UsageError:  # login anonymously if not logged in already
-    wandb.login(
-        host="https://genentech.wandb.io", relogin=True, anonymous="must", timeout=0
-    )
+    wandb.login(host=DEFAULT_WANDB_HOST, relogin=True, anonymous="must", timeout=0)
 
 
 # Test a fully convolutional model with residual connections and autocropping