From: Jérôme Benoit <jerome.benoit@piment-noir.org>
Date: Sat, 31 Jan 2026 20:48:50 +0000 (+0100)
Subject: fix(ReforceXY): use .item() to extract scalar from action array
X-Git-Url: https://git.piment-noir.org/?a=commitdiff_plain;h=123843f1848335800298d3eb15c7e78c4622c347;p=freqai-strategies.git

fix(ReforceXY): use .item() to extract scalar from action array

SB3 model.predict() returns 1D array when vectorized_env=True is detected, causing int() to fail on non-0D arrays.
---

diff --git a/ReforceXY/user_data/freqaimodels/ReforceXY.py b/ReforceXY/user_data/freqaimodels/ReforceXY.py
index 16d185e..174a78e 100644
--- a/ReforceXY/user_data/freqaimodels/ReforceXY.py
+++ b/ReforceXY/user_data/freqaimodels/ReforceXY.py
@@ -1177,7 +1177,8 @@ class ReforceXY(BaseReinforcementLearningModel):
                     **action_masks_param,
                 )
                 episode_start[:] = False
-                logger.debug("Predict [%s]: predicted action=%d", dk.pair, int(action))
+                action = int(action.item())
+                logger.debug("Predict [%s]: predicted action=%d", dk.pair, action)
             else:
                 logger.debug(
                     "Predict [%s]: model.predict (observations.shape=%s)",
@@ -1187,9 +1188,10 @@ class ReforceXY(BaseReinforcementLearningModel):
                 action, _ = model.predict(
                     observations, deterministic=True, **action_masks_param
                 )
-                logger.debug("Predict [%s]: predicted action=%d", dk.pair, int(action))
+                action = int(action.item())
+                logger.debug("Predict [%s]: predicted action=%d", dk.pair, action)
 
-            return int(action)
+            return action
 
         predicted_actions: List[int] = []
         for start_idx in range(0, n - window_size + 1):