Evaluation Metrics for the default model weights

[RamvigneshPasupathy]

In addition to the Benchmark info provided in the README of this repo, it will be highly helpful and good to know what was the training and evaluation loss / accuracy for the default model weights.