Refined review

src/brevitas_examples/imagenet_classification/ptq/benchmark/ptq_benchmark_torchvision.py

@@ -131,12 +131,21 @@ def main():
 
 def ptq_torchvision_models(df, args):
     # Generate all possible combinations, including invalid ones
-    combinations = list(product(*OPTIONS.values()))
+    # Split stats and mse due to the act_quant_percentile value
+    percentile_options = OPTIONS.copy()
+    percentile_options['act_param_method'] = ['stats']
+    mse_options = OPTIONS.copy()
+    mse_options['act_param_method'] = ['mse']
+    mse_options['act_quant_percentile'] = [None]
+    # Combine the two sets of combinations
+    combinations = list(product(*percentile_options.values())) + list(
+        product(*mse_options.values()))
+    # Generate Namespace for each configuration
     configs = [
         SimpleNamespace(**{k: v
                            for k, v in zip(OPTIONS.keys(), combination)})
         for combination in combinations]
-    # Define which configs are not valid
+    # Define which configurations are not valid
     configs = list(map(validate_config, configs))
     # Drop invalid configurations
     configs = list(config for config in configs if config.is_valid)
@@ -280,11 +289,6 @@ def validate_config(config_namespace):
         is_valid = False
     if config_namespace.act_bit_width < config_namespace.weight_bit_width:
         is_valid = False
-    if config_namespace.act_param_method == 'mse':
-        if config_namespace.act_quant_percentile == 99.999:
-            config_namespace.act_quant_percentile = None
-        else:
-            is_valid = False
 
     config_namespace.is_valid = is_valid
     return config_namespace