Val Storm

I've trained a model to 100k steps on the GPU When using the generate.py script my gpu is listed, CUDA opens successfully but samples are generated at only 200-400 samples...

From Max I'm sending "/beats stream 17005" Then this in the console ``` java.nio.channels.AsynchronousCloseException at java.base/java.nio.channels.spi.AbstractInterruptibleChannel.end(AbstractInterruptibleChannel.java:202) at java.base/sun.nio.ch.DatagramChannelImpl.endWrite(DatagramChannelImpl.java:1030) at java.base/sun.nio.ch.DatagramChannelImpl.send(DatagramChannelImpl.java:827) at overtone.osc.peer$chan_send.invokeStatic(peer.clj:224) at overtone.osc.peer$chan_send.invoke(peer.clj:216) at overtone.osc.peer$send_loop$fn__938.invoke(peer.clj:107) at overtone.osc.peer$send_loop.invokeStatic(peer.clj:102) at overtone.osc.peer$send_loop.invoke(peer.clj:84)...