Visual watcher when training/evaluating or tensorboard equivalence?

8

Nevermind, it works without the extra lines,

foo.get_scalar_values("Mean Reward")
foo.get_scalar_values("Max Reward")

Also I needed to give distinct date time’s for each of the threads,

cc = CrayonClient( hostname="http://127.0.1.1" , port=8889)
exp_name = "train_" + str(rank) + "_" + datetime.now().strftime('train_%m-%d_%H-%M') # prevents server errors ??
foo = cc.create_experiment(exp_name)
1 Like