microrts-ai trained against randomBiasedAI

Created on April 4|Last edited on April 4
Comment
﻿
unet-MicrortsDefeatRandomEnemySparseReward-v3
benchmark_388c8ed host_155-248-197-5 branch_unet v0.0.8
﻿https://github.com/sgoodfriend/rl-algo-impls/commit/388c8ed﻿
Eval using different scoring function, 4000 steps, and gamma 0.999
benchmark_44dc2ec host_192-9-147-179 branch_unet v0.0.8
﻿https://github.com/sgoodfriend/rl-algo-impls/commit/44dc2ec﻿
Progressive reward decay
benchmark_ab60ce1 host_150-230-47-100 branch_progressive_reward_callback v0.0.8
﻿https://github.com/sgoodfriend/rl-algo-impls/commit/ab60ce1﻿
unet-MicrortsDefeatCoacAIShaped-v3-diverseBots﻿
train_rolling/microrts_stats_WinLossRewardFunction
train_rolling/microrts_stats_WinLossRewardFunction
500k1M1.5Mglobal_step-0.500.5
algo_hyperparams.gamma: -, microrts_reward_decay_callback: -
algo_hyperparams.gamma: 0.999, microrts_reward_decay_callback: -
algo_hyperparams.gamma: 0.999, microrts_reward_decay_callback: true
Run set9
﻿
﻿
﻿
﻿
﻿
﻿﻿﻿
﻿
﻿
Add a comment