-
Of course that 20 block net has the fixed limit.
But I don't think that they moved to 40 block due to the limit of network.
(This opinion became popular after Hideki Kato which is developer of DeepZ…
-
Business/invention ideas listed here for publication.
-
- Abstract (2-3 lines)
Neural Style Transfer is a technique using which one can convert there own pictures into amazing artistic pieces! It uses one Content Image and one Style Image (which is usuall…
-
- [x] Add convolutional net to the softmax IJulia. (done in dl-tutorial/cnn.ipynb)
- [x] Fix RNN to use the more efficient indexing approach in rnn IJulia and charlm example. (switched to cudnn inter…
-
Creating this topic because otherwise we'll see 10 identical issues pointing out the obvious.
It looks like 6x128 will have its last learning rate lowering soon. I've been training up 10x128 starti…
-
Hello,
I did not find if anyone tried this TF version of the ladder network fully supervised to compare with the paper's baseline MLP result from table 1 so I tested it for that case.
I am running …
-
A while ago I realized that [the color planes can be used to input komi information to the network](https://github.com/gcp/leela-zero/issues/1576); today I finally have some time to implement the recu…
-
Hi,
I'm curious about how other experiments performed, it would be great to add these results to the readme if possible.
For example, have you tried using the entire image as input instead of small …
-
all of the networks after lz226 have winrates less than 0.45
-
# Problem description
TS 26.501 defines Downlink Network Assistance(NA) as a feature that _"enables a UE that is receiving a downlink media stream to improve the QoE of the media streaming session, b…