dlbench
dlbench copied to clipboard
hclhkbu
Benchmarking State-of-the-Art Deep Learning Software Tools
From benchmark.py and configs/*.config, we know dlbench provide capability of changing learning rate. However, only Caffe, Torch, MXNet accepts learning rate argument while CNTK, TensorFlow ignores them. ``` # tools/cntk/cntkbm.py...
Judging from http://dlbench.comp.hkbu.edu.hk/logs/?f=tensorflow-rnn-lstm-gpu0-K80-b1024-Wed_Mar__8_17:32:12_2017-gpu20.log LSTM code isn't updated for TF 1.0
Hi! First of all, this is the best benchmark that i've seen (that includes both cnn and rnn). Are you planning to include the pytorch in the future? Thanks.
I'm Tensorflow user, so I've opened [an issue](https://github.com/tensorflow/tensorflow/issues/7187) regarding performance of Tensorflow in several cases. One of things we found out is that the code used by dlbench is suboptimal...
Hi, The versions of mxnet and tensorflow are too old. Could you update these tools and share the results. I tried to run the benchmark however it seems the code...
Hi, Really appreciate the work put in by you guys to compare all these libraries across so many different models and configurations. I had a minor nitpick. For Torch7, it...
Theano?
Thanks for the great benchmark! Are you considering to include Theano in the future?
It would be useful to have a readme to reproduce the benchmarks. For instance I found that for running cifar multi-gpu resnet on TF I needed to do `pip install...
