Giters

google-deepmind / open_spiel

OpenSpiel is a collection of environments and algorithms for research in general reinforcement learning and search/planning in games.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

chat_game_base.py prints lots of stuff during testing

tacertain opened this issue · comments

commented

When open_spiel/integration_tests/playthrough_test.py is run, most of the tests run silently, giving output such as:

[ RUN      ] PlaythroughTest.test_playthrough_pig_3p.txt
[       OK ] PlaythroughTest.test_playthrough_pig_3p.txt
[ RUN      ] PlaythroughTest.test_playthrough_pig_4p.txt
[       OK ] PlaythroughTest.test_playthrough_pig_4p.txt
[ RUN      ] PlaythroughTest.test_playthrough_pig_5p.txt
[       OK ] PlaythroughTest.test_playthrough_pig_5p.txt
[ RUN      ] PlaythroughTest.test_playthrough_python_block_dominoes.txt
[       OK ] PlaythroughTest.test_playthrough_python_block_dominoes.txt

However, chat_game_base.py prints lots and lots of output, such as:

I0514 19:03:37.891809 139833831002112 chat_game_base.py:438] Extracting payoff 0 (player=1:Suzy)...
I0514 19:03:37.891947 139833831002112 chat_game_base.py:444] LLM prompt:
Extract out the final value for Suzy as a single numeric value from the following payoff valuation. Do NOT show your work:

5


Result:
I0514 19:03:37.892134 139833831002112 chat_game_base.py:450] LLM response:
5

I0514 19:03:37.892304 139833831002112 chat_game_base.py:460] Extracted integer payoff (Suzy): 5
I0514 19:03:37.892467 139833831002112 chat_game_base.py:428] Calculating payoff 1 (player=1:Suzy)...

It makes it much harder to scan through the results for things going poorly.