| 247 | return self |
| 248 | |
| 249 | def _get_init_kwargs(self): |
| 250 | init_kwargs = dict( |
| 251 | learning_rate=self.learning_rate, |
| 252 | n_steps_total=self.n_steps_total, |
| 253 | batch_size=self.batch_size, |
| 254 | n_epochs=self.n_epochs, |
| 255 | gamma=self.gamma, |
| 256 | gae_lambda=self.gae_lambda, |
| 257 | clip_range=self.clip_range, |
| 258 | clip_range_vf=self.clip_range_vf, |
| 259 | ent_coef=self.ent_coef, |
| 260 | explore_coef=self.explore_coef, |
| 261 | vf_coef=self.vf_coef, |
| 262 | max_grad_norm=self.max_grad_norm, |
| 263 | target_kl=self.target_kl, |
| 264 | update_adv=self.update_adv, |
| 265 | lr_schedule_step=self.lr_schedule_step, |
| 266 | start_num_timesteps=self.num_timesteps, |
| 267 | ) |
| 268 | return init_kwargs |
| 269 | |
| 270 | def save(self, path: str) -> None: |
| 271 | th.save({'policy_state_dict': self.policy.state_dict(), |