Player process 1.0

[justheuristic]

A process that

• compiles the agent
• takes the stored dqn params every once in a while
  • if there are no params, put them there
• interacts :) (cpu)
• stores the interactions in the database
• all interactions are of same length (e.g. 10 ticks)
• loads new weights every once in a while
• does not break down everything if restarted at random point of time
• does not erase old sessions
• does not intentionally break down if there are several such processes running in parallel :)

Basically do that but with an agent and all the trouble that comes with it. Example agent setup for mountaincar.

[justheuristic]

Okay, so we gonna have experiment description like this minimal example and hopefully somewhat less crutchy.

There's some skeleton of the base player process you may want to start from:

• https://github.com/yandexdataschool/tinyverse/blob/master/player.py
• how to run it
• see the gpu server location in gitter

Currently it only plays the base game with current weights, does not try to load new ones, does not log anything and probably may become much quicker.

I suggest starting with loading new weights once in a while (you may need a parameter for this).

Btw pls write something if you're still there :)

[justheuristic]

Merged https://github.com/yandexdataschool/tinyverse/commit/c92a8d0bce48a7c33aa74633028138fc9059d290