-
### _TFX RFP 001 - Request for community comments and proposals_
# Adding Non-Transform Feature Engineering
_An explanation of a specific critical user journey (CUJ) for windowing, and a general…
-
## Motivation
Machine learning models are generally black boxes. These are difficult to explain or get to know the internal workings of. Meanwhile, they often provide state of the art performance in …
-
Following offline discussion with @lostanlen. Since examples are not tested regularly, they *will* break (see #817, #801, #775) without us realizing it. We should aim to have the examples run at somew…
-
My last layer uses tanh activation function for continuous variable prediction like steering angle. What changes I need in deeplift to use deeplift for my scenario i.e., regression problem.
-
The CRAFT authors used a weakly-supervised training method to handle the fact that most datasets don't annotate at the character level. I saw in your [docs](https://keras-ocr.readthedocs.io/en/latest/…
-
Hi,
First of all, thanks for publicly sharing your implementations of the reinforcement learning algorithms. I find your repos very useful!
As I was playing around with the QR-DQN, I think I not…
-
### System information
Linux OpenSuse Tumbleweed
- TensorFlow installation : pip
- TensorFlow library : Tf-nightly, occurs on earlier versions too
### Code
Converting a model containing an …
-
Is it possible to use `Keras tuner` for tuning a NN using `Time Series Split` , similar to `sklearn.model_selection.TimeSeriesSplit` in `sklearn`.
For example consider a sample tuner class from [ht…
-
UPD:
The standing items:
- add pseudocode. I think pseudocode / compact python implementations would be very useful for understanding what is normalized over and what learned affine weights corres…
-
Hi, I want to convert weights to tflite using tflite_convert --keras_model_file=yolov3-tiny.h5 --output_file=yolov3-tiny.tflite
It fails with the following traceback:
Instructions for updating:
C…