updated examples to use new estimators

joshuaspear · Jul 17, 2024 · a1a86b0 · a1a86b0
1 parent 63e0775
commit a1a86b0
Show file tree

Hide file tree

Showing 3 changed files with 12 additions and 12 deletions.
diff --git a/examples/static_torch_deterministic_continuous.py b/examples/static_torch_deterministic_continuous.py
@@ -17,7 +17,7 @@
 from offline_rl_ope.components.Policy import Policy, GreedyDeterministic
 from offline_rl_ope.components.ImportanceSampler import ISWeightOrchestrator
 from offline_rl_ope.OPEEstimators import (
-    ISEstimator, DREstimator, D3rlpyQlearnDM)
+    ISEstimator, WDR, D3rlpyQlearnDM)
 from offline_rl_ope.PropensityModels.torch import FullGuassian, TorchRegTrainer 
 from offline_rl_ope.LowerBounds.HCOPE import get_lower_bound
 
@@ -176,9 +176,9 @@ def __call__(
     wis_estimator_smooth = ISEstimator(norm_weights=True, norm_kwargs={
         "smooth_eps":0.0000001
     })
-    w_dr_estimator = DREstimator(
-        dm_model=fqe_dm_model, norm_weights=True, 
-                                ignore_nan=True)
+    w_dr_estimator = WDR(
+        dm_model=fqe_dm_model, 
+        )
 
 
     res = is_estimator.predict(

diff --git a/examples/static_torch_stochastic_continuous.py b/examples/static_torch_stochastic_continuous.py
@@ -17,7 +17,7 @@
 from offline_rl_ope.components.Policy import Policy
 from offline_rl_ope.components.ImportanceSampler import ISWeightOrchestrator
 from offline_rl_ope.OPEEstimators import (
-    ISEstimator, DREstimator, D3rlpyQlearnDM)
+    ISEstimator, WDR, D3rlpyQlearnDM)
 from offline_rl_ope.PropensityModels.torch import FullGuassian, TorchRegTrainer 
 from offline_rl_ope.LowerBounds.HCOPE import get_lower_bound
 
@@ -158,9 +158,9 @@
     wis_estimator_smooth = ISEstimator(norm_weights=True, norm_kwargs={
         "smooth_eps":0.0000001
     })
-    w_dr_estimator = DREstimator(
-        dm_model=fqe_dm_model, norm_weights=True, 
-                                ignore_nan=True)
+    w_dr_estimator = WDR(
+        dm_model=fqe_dm_model
+        )
 
 
     res = is_estimator.predict(

diff --git a/examples/static_xgboost_discrete.py b/examples/static_xgboost_discrete.py
@@ -16,7 +16,7 @@
     GreedyDeterministic, Policy, NumpyPolicyFuncWrapper)
 from offline_rl_ope.components.ImportanceSampler import ISWeightOrchestrator
 from offline_rl_ope.OPEEstimators import (
-    ISEstimator, DREstimator, D3rlpyQlearnDM)
+    ISEstimator, WDR, D3rlpyQlearnDM)
 from offline_rl_ope.PropensityModels.sklearn import (
     SklearnDiscrete)
 from offline_rl_ope.LowerBounds.HCOPE import get_lower_bound
@@ -123,9 +123,9 @@
     wis_estimator_smooth = ISEstimator(norm_weights=True, norm_kwargs={
         "smooth_eps":0.0000001
     })
-    w_dr_estimator = DREstimator(
-        dm_model=fqe_dm_model, norm_weights=True, 
-                                ignore_nan=True)
+    w_dr_estimator = WDR(
+        dm_model=fqe_dm_model
+        )
 
 
     res = is_estimator.predict(