[#130]Debug/fix step calculation in async_distributed_train

:star2: Hello! Thanks for contributing JORLDY!

Checklist

Please check if you consider the following items.

[x] My code follows the style guidelines of this project contributing
[x] My code follows the naming convention of documentation
[x] I have commented my code, particularly in hard-to-understand areas
[x] My changes generate no new warnings or errors

Types of changes

Please describe the types of changes! (ex. Bugfix, New feature, Documentation, ...) Bugfix

Test Configuration

OS: mac
Python version: 3.8.12
Additional libraries: vscode-conda

Description

Please describe the details of your contribution

In each asynchronous loop, a different number of steps comes in each time, and now it is reset to "heap["print_stamp"] = 0 " as in the following code.

if heap["print_stamp"] >= config.train.print_period or is_over:
    print_signal = True
    heap["print_stamp"] = 0 # 50100.25->0, loss 100.25 step

...

if print_signal: # 49899.75 >= 50000 at last loop
    try:
        manage_sync_queue.get_nowait()
    except:
        pass
    manage_sync_queue.put(agent.sync_out()) # no execution in the last loop.

In fact, if it is updated at heap["print_stamp"]=50100 of 50000 period, it should be updated after an additional 49900 steps. Currently, while updating heap["print_stamp"]=0, the entire step is finished after the last additional 49900 steps, but a is not updated and the loop does not end.

So, fix it like this:

if heap["print_stamp"] >= config.train.print_period or is_over:
    print_signal = True
    heap["print_stamp"] -= config.train.print_period # fix

kakaoenterprise / JORLDY