(env, pmodel, gamma)
| 209 | |
| 210 | |
| 211 | def random_search(env, pmodel, gamma): |
| 212 | totalrewards = [] |
| 213 | best_avg_totalreward = float('-inf') |
| 214 | best_pmodel = pmodel |
| 215 | num_episodes_per_param_test = 3 |
| 216 | for t in range(100): |
| 217 | tmp_pmodel = best_pmodel.copy() |
| 218 | |
| 219 | tmp_pmodel.perturb_params() |
| 220 | |
| 221 | avg_totalrewards = play_multiple_episodes( |
| 222 | env, |
| 223 | num_episodes_per_param_test, |
| 224 | tmp_pmodel, |
| 225 | gamma |
| 226 | ) |
| 227 | totalrewards.append(avg_totalrewards) |
| 228 | |
| 229 | if avg_totalrewards > best_avg_totalreward: |
| 230 | best_pmodel = tmp_pmodel |
| 231 | best_avg_totalreward = avg_totalrewards |
| 232 | return totalrewards, best_pmodel |
| 233 | |
| 234 | |
| 235 | def main(): |
no test coverage detected