Reproducing Figure 1 of DAWNBench with TensorFlow

This notebook reproduces Figure 1 of the DAWNBench experiment with Tensorflow. DAWNBench is a project at Stanford University that introduces a benchmark and competition that focused on end-to-end training time for a model to reach a fixed accuracy level. The paper is available at https://dawn.cs.stanford.edu/benchmark/papers/nips17-dawnbench.pdf. Figure 1 (the graph that we reproduce here) shows the relationship between the accuracy level and the end-to-end training time for three different batch sizes.