Luca Urbinati

I add a contribution to @DirtyEngineer's answers. I used his code to plot a full screen 128x32 image on my OLED. Here is my code, maybe someone can be interested...

I have the same problem. I think that this line should be changed from this: https://github.com/keras-team/autokeras/blob/0d22c6f6a611cdfc017a24c28c68e7925b7f7feb/autokeras/engine/tuner.py#L63 to something like this: `model = tf.keras.models.load_model(self.best_model_path, custom_objects={"custom_metric": custom_metric}) ` when there is a...

I thought I had the same problem of @rdp23, but then I realized that I wasn't calculating well the quantized values that I was using as reference for the comparison...

Read my questions and answers in this [topic](https://github.com/hyperopt/hyperopt/issues/267). In short, I changed all my code to Hyperopt (thanks to this [examples](https://www.programcreek.com/python/example/98788/hyperopt.Trials)).

Hi, is there the possibility to effectively produce a quantized model from a QAT model? I read this "[converting from a quantization-aware training model to a post-training quantization model is...

Hi @LucasFischer123, **Short answer** We "solved" it by using 10% of the whole [dataset](https://github.com/mlcommons/tiny/blob/master/benchmark/training/visual_wake_words/download_and_train_vww.sh) as "validation set" during training (according to the [train_vww.py](https://github.com/mlcommons/tiny/blob/master/benchmark/training/visual_wake_words/train_vww.py) script) and then using [these 1000 images](https://github.com/mlcommons/tiny/blob/master/benchmark/evaluation/datasets/vww01/y_labels.csv)...

I have a question for @bjkomer and @cbaziotis because I implemented your codes. When the number of iterations arrives to be around 50, loading and saving with pickle becomes too...