Nick Meehan

I don't think this would work or apply to a sequential batch input if I'm not mistaken. Is there a solution for predict_on_batch in that instance?

Where the data is sequential time based data that is trained in batches where sequencing and "statefulness" matters (think large char-rnn or word-rnn).

I'm relatively new to keras so forgive me if I have something confused. Btw y'all have been on a roll squashing bugs today!

Have a simpler solution, though yes upgrading cmake works, adding a flag works as well. ``` export CUDA_CUDART_LIBRARY="/usr/local/cuda/lib64/libcudart.so" cd nheqminer/cpu_xenoncat/asm_linux \ && sh assemble.sh \ && cd /home/nheqminer \ &&...

I put together a docker image that seems to do the job on ubuntu, and it runs CUDA --- covers most of the above issues. https://hub.docker.com/r/unsalted/nheqminer/

Check your cmake version. On Apr 12, 2017, at 2:36 PM, napoleon2121 wrote: make[2]: *** No rule to make target '../../cpu_xenoncat/Linux/asm/equihash_avx1.o', needed by 'nheqminer_cpu'. Stop. CMakeFiles/Makefile2:67: recipe for target 'CMakeFiles/nheqminer_cpu.dir/all'...

That looks correct -- here is my dockerfile, make sure boost is the right version as well. https://github.com/unsalted/docker-nheqminer-cuda/blob/master/Dockerfile On Apr 12, 2017, at 6:58 PM, napoleon2121 wrote: cmake version 3.5.2...

Same on 1x GTX 1070 -- after about 2hrs

Don't know if it is related but the custom script solved this for me, I needed to add routing and parent when indexing a child document and this fixed the...

Would love to see this merged in and officially supported. @melishev