David Ding, Lily Tsai, Serena Wang
##Evaluation
- Agents vs. Benchmarks:
- Top agent-trained agents A0-A4
- Benchmark-trained agents B0-B4
- Initial agents in the agent-training population A5-A9
- Initial agents in the benchmark-training population against B5-B9
./evaluation_scripts/run_all_against_benchmarks.sh
- Top Agents in Agent-Trained Population vs. Top Agents in Benchmark-Trained Population:
./evaluation_scripts/agent_vs_benchmark_training_full_population.sh
- Top Agent-Trained (Coevolved) Agents vs. Top Benchmark-Trained Agents:
./evaluation_scripts/coevolution_vs_evolution.sh
- Top Initial Agents in Agent-Training Population vs. Top Agent-Trained Agents:
./evaluation_scripts/initial_agent_trained_vs_trained.sh
- Top Initial Agents in Benchmark-Training Population vs. Top Benchmark-Trained Agents:
./evaluation_scripts/initial_agent_trained_vs_trained.sh
##Evolution
Usage: python evolution/evolution_agent.py
[-h] [--bmfile BMFILE]
[--nn_agent_file NN_AGENT_FILE]
[--agent_param_dir AGENT_DIR] [--gamefile GAMEFILE]
[--ntopagents NTOPAGENTS] [--epochs EPOCHS]
[--nagents NAGENTS]
[--nagents_to_mutate NAGENTS_TO_MUTATE]
[--nagents_to_keep NAGENTS_TO_KEEP] [--coevolve]
[--nthreads NTHREADS] [--nhands NHANDS]