added weight standard deviation metric

src/offline_rl_ope/Metrics/MetricBase.py

@@ -11,4 +11,4 @@ class MetricBase(metaclass=ABCMeta):
 
     @abstractmethod
     def __call__(self, weights:torch.Tensor, *args:Any, **kwargs:Any) -> float:
-        return self.__ess(weights=weights)
+        pass

src/offline_rl_ope/Metrics/ValidWeightsProp.py

@@ -1,6 +1,5 @@
 import torch
 from .MetricBase import MetricBase
-from ..OPEEstimators.utils import get_traj_weight_final
 from jaxtyping import jaxtyped
 from typeguard import typechecked as typechecker
 

src/offline_rl_ope/Metrics/WeightStd.py

@@ -0,0 +1,21 @@
+import torch
+from .MetricBase import MetricBase
+from jaxtyping import jaxtyped
+from typeguard import typechecked as typechecker
+
+from ..types import WeightTensor
+
+__all__ = [
+    "WeightStd"
+]
+
+class WeightStd(MetricBase): 
+
+    @jaxtyped(typechecker=typechecker)
+    def __call__(
+        self, 
+        weights:WeightTensor, 
+        weight_msk:WeightTensor
+        ) -> float:
+        sum_weights = torch.mul(weights,weight_msk).sum(dim=1)
+        return torch.std(sum_weights).item()

src/offline_rl_ope/Metrics/__init__.py

@@ -1,3 +1,4 @@
 from .EffectiveSampleSize import *
 from .ValidWeightsProp import *
+from .WeightStd import *
 from .MetricBase import * 

tests/Metrics/test_ValidWeightsProp.py

@@ -17,7 +17,7 @@ def test_call(self):
         min_val=0.000001
         fnl_weights = []
         for idx,i in enumerate(self.test_conf.traj_lengths):
-            fnl_weights.append(self.test_conf.weight_test_res[idx,:i-1].sum(
+            fnl_weights.append(self.test_conf.weight_test_res[idx,:i].sum(
                 dim=0,
                 keepdim=True
                 ))

tests/Metrics/test_WeightStd.py

@@ -0,0 +1,31 @@
+import unittest
+import torch
+import numpy as np
+import copy
+from offline_rl_ope.Metrics import WeightStd
+from offline_rl_ope import logger
+from parameterized import parameterized_class
+from ..base import test_configs_fmt_class, TestConfig
+
+@parameterized_class(test_configs_fmt_class)
+class TestWeightStd(unittest.TestCase):
+
+    test_conf:TestConfig
+
+    def test_call(self):
+        fnl_weights = []
+        for idx,i in enumerate(self.test_conf.traj_lengths):
+            fnl_weights.append(
+                self.test_conf.weight_test_res[idx,:i].sum(
+                    dim=0,
+                    keepdim=True
+                    )
+                )
+        fnl_weights_tens = torch.concat(fnl_weights, axis=0)
+        act_res = torch.std(fnl_weights_tens).item()
+        metric = WeightStd()
+        pred_res = metric(
+            weights=self.test_conf.weight_test_res, 
+            weight_msk=self.test_conf.msk_test_res
+            )
+        self.assertEqual(act_res,pred_res)