-
Hi, I'm interested in your HTM Spatial Pooler and the paper in the title.
Here's the question: the `SparseNet` introduced in the paper generates sparse _scalar_ vectors. But the original HTM genera…
-
When I use keras LSTM for part-of-speech tagging, I used `mask_value=0` to mask the 0 value timesteps in the input, and it worked because the output prob are all same in the input end, for example, in…
-
This is a list of things that I noticed while working with some learners. I'm no expert on any of the learners, but the documentation or my intuition told me these might be bugs. Maybe some of these t…
mb706 updated
4 years ago
-
like [tf.keras.layers.UpSampling2D](https://www.tensorflow.org/api_docs/python/tf/keras/layers/UpSampling2D)
even just "nearest" method. this is useful for some image restoration problem like denoise…
-
This issue collects resources for the newsletter of the year/month/week that should be in the title
If you want to contribute open the issue and paste a resource you think is worth sharing in the n…
-
#### Issue description
While working with #2048, I noticed that if `rho` is longer than the sequence length, an error is given:
```
error: Cube::slice(): index out of bounds
terminate ca…
-
strange compile error using double:
collect2.exe: error: ld returned 1 exit status
does not happen with float vs. double
Basic Infos
[x ] This issue complies with the issue POLICY doc.
…
-
According to the Forest docs, the "new way" of minimizing an anzatz does the following:
1. declares parameters in the program, and
2. uses the memory_map dict as an argument to qc.run().
But you…
-
Hi!
I want to use MobileNetV2 instead of DarkNet,
But I can't get accuracy more than 0.003 after 10 hours of training.
I'm training the network on the WIDER FACE dataset.
total data is 11753
…
-
Hello,
This is probably not a bug nor an issue, but rather a question about the possible reasons for the behaviour of the CG method of scipy.optimize.minimize.
I built a neural network with a si…