(self, state, action, reward, next_state, done)
| 268 | |
| 269 | |
| 270 | def train(self, state, action, reward, next_state, done): |
| 271 | if done: |
| 272 | target = reward |
| 273 | else: |
| 274 | target = reward + self.gamma * np.amax(self.model.predict(next_state), axis=1) |
| 275 | |
| 276 | target_full = self.model.predict(state) |
| 277 | target_full[0, action] = target |
| 278 | |
| 279 | # Run one training step |
| 280 | self.model.sgd(state, target_full) |
| 281 | |
| 282 | if self.epsilon > self.epsilon_min: |
| 283 | self.epsilon *= self.epsilon_decay |
| 284 | |
| 285 | |
| 286 | def load(self, name): |
no test coverage detected