Disable per-channel quant on requantize test (#2616)

BUG=cleanup
tensorflow · Jul 2, 2024 · 0cc2e49 · 0cc2e49
1 parent dc4dcb7
commit 0cc2e49
Showing 1 changed file with 1 addition and 0 deletions.
diff --git a/tensorflow/lite/micro/tools/requantize_flatbuffer_test.py b/tensorflow/lite/micro/tools/requantize_flatbuffer_test.py
@@ -62,6 +62,7 @@ def convert_tfl_converter(keras_model,
   converter.representative_dataset = representative_dataset_gen
   # TODO(b/324385802): Support per-channel quantization for FullyConnected.
   converter._experimental_disable_per_channel_quantization_for_dense_layers = True
+  converter._experimental_disable_per_channel = True
   return converter.convert()