N
N
@UsharaniPagadala as you can see already in the code, I stumbled acorss both of these issues previously. I am normalizing accordingly (range 0 - 255 for keras, 0 - 1...
Copying the tutorial code leads to exactly the same (erroneous) outputs as in my example; see code snippet below. The reason WHY I opened this ticket was because I saw...
Hi @maringeo , as I already stated above I am aware that the rescaling needs to be done prior to feeding the image to tfhub as per the image conventions....
I tried checking in the code which is supposed to be used during training and I can find the following traces: https://github.com/tensorflow/models/blob/30e6e03f66efad4e43f1b98ec8680451f5a86a72/official/vision/image_classification/preprocessing.py#L168 Note that they essentially call ```python MEAN_RGB =...
Hi @saberkun ,, Thanks for clarifying this. To summarize my understanding: tfhub/model-garden expects the input to be [0...1], which then gets converted to [0...255] as a first step, and then...
I will try to find some time to provide a side by side example/code for training showcasing the performance changes; however this may take a while due to work priorities.
I managed to (somewhat!) reproduce the performance gap (Note the differences in loss and val_loss between the two...) on the tf flower dataset below. Interestingly, the hub model trains a...
I have no idea what "Day of week of month / F" is supposed to contain. The rest I should be able to implement in the attached PR.
If thats fine, I'll have a look into it. My idea for now would be to keep syntax in line with the approach used by row_number, i.e. lag(column, N), or...
https://github.com/questdb/questdb/blob/master/core/src/main/java/io/questdb/griffin/engine/groupby/GroupByUtils.java#L79 this is the assertion that is usually tripped, because https://github.com/questdb/questdb/blob/master/core/src/main/java/io/questdb/griffin/FunctionParser.java#L248 attempts converting functions to constants for performance reasons. So I think that - We need to make sure that...