popov/gogo2
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fixes

Browse Source
This commit is contained in:
Dobromir Popov
2025-06-24 21:25:20 +03:00
parent e7ea17b626
commit f47cf52ae1
1 changed files with 15 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
core/enhanced_orchestrator.py
View File

@ -1395,13 +1395,14 @@ class EnhancedTradingOrchestrator(TradingOrchestrator):
predictions = prediction_result.get('probabilities', [0.33, 0.33, 0.34])
confidence = prediction_result.get('confidence', 0.7)
# Convert predictions to numpy array first
# Convert predictions to numpy array first using safe conversion
if isinstance(predictions, np.ndarray):
predictions_array = predictions.flatten()
elif isinstance(predictions, (list, tuple)):
predictions_array = np.array(predictions, dtype=np.float32).flatten()
else:
predictions_array = np.array([float(predictions)], dtype=np.float32)
# Use safe tensor conversion for single values
predictions_array = np.array([self._safe_tensor_to_scalar(predictions, 0.5)], dtype=np.float32)
# Create final predictions array with confidence
# Use safe tensor conversion to avoid scalar conversion errors
@ -1416,7 +1417,7 @@ class EnhancedTradingOrchestrator(TradingOrchestrator):
# Handle (pred_class, pred_proba) tuple from CNN models
pred_class, pred_proba = prediction_result
# Flatten and process the probability array
# Flatten and process the probability array using safe conversion
if isinstance(pred_proba, np.ndarray):
if pred_proba.ndim > 1:
# Handle 2D arrays like [[0.1, 0.2, 0.7]]
@ -1428,16 +1429,17 @@ class EnhancedTradingOrchestrator(TradingOrchestrator):
# Use the probability values as the predictions array
predictions = pred_proba_flat.astype(np.float32)
else:
# Fallback: use class prediction only
predictions = np.array([float(pred_class)], dtype=np.float32)
# Fallback: use class prediction with safe conversion
predictions = np.array([self._safe_tensor_to_scalar(pred_class, 0.5)], dtype=np.float32)
else:
# Handle direct prediction result
# Handle direct prediction result using safe conversion
if isinstance(prediction_result, np.ndarray):
predictions = prediction_result.flatten()
elif isinstance(prediction_result, (list, tuple)):
predictions = np.array(prediction_result, dtype=np.float32).flatten()
else:
predictions = np.array([float(prediction_result)], dtype=np.float32)
# Use safe tensor conversion for single tensor/scalar values
predictions = np.array([self._safe_tensor_to_scalar(prediction_result, 0.5)], dtype=np.float32)
# Extract hidden features if model supports it
hidden_features = None
@ -4740,7 +4742,8 @@ class EnhancedTradingOrchestrator(TradingOrchestrator):
Python float scalar value
"""
try:
if hasattr(tensor_value, 'item'):
# Handle PyTorch tensors first
if hasattr(tensor_value, 'numel') and hasattr(tensor_value, 'item'):
# PyTorch tensor - handle different shapes
if tensor_value.numel() == 1:
return float(tensor_value.item())
@ -4754,9 +4757,12 @@ class EnhancedTradingOrchestrator(TradingOrchestrator):
return float(tensor_value.flatten()[0])
else:
return float(tensor_value.flat[0])
elif hasattr(tensor_value, 'item') and not isinstance(tensor_value, np.ndarray):
# Other tensor types that have .item() method
return float(tensor_value.item())
else:
# Already a scalar value
return float(tensor_value)
except Exception as e:
logger.warning(f"Error converting tensor to scalar, using default {default_value}: {e}")
logger.debug(f"Error converting tensor to scalar, using default {default_value}: {e}")
return default_value