LUX S2 Submission Analysis
Created on April 25|Last edited on April 30
Comment
Submission Run
A disappointing follow-up run
After the excellent 1c3f35f4 run, I had hopes that I could get excellent follow-up results:
b4b9810 close_extras takes arguments that have to be handled66ed1e5 Small and medium use same map size for eval63681a9 Fix to support different eval map vs train map2c371c1 Fix imports2684f03 mean score_function for eval5f3766b Move Lux StatsTracking into a lux/stats module4f3b153 Eval only tracks win/loss11149af factories_alive stat awards current number of factoriesbbfa3ee medium-debug env hyperparams for local133c8eb medium 30M steps (should complete in 1 day)f9d6769 LuxAI_S2-v0-medium nearly guarantees existence of iced70c4ac Tweak small to not print as much and debug70c8be3 LuxAI_S2-v0-small is a 16x16, 1 factory map
The only apparent difference in behavior is the FACTORIES_ALIVE reward weighed at 0.02
Add a comment