jumpstart-rl

jumpstart-rl copied to clipboard

Thank you for the great work! I am wondering how to set the initial exploration as the guided policy rather than from scratch. Any ideas?

chengzelei

unexpected keyword argument 'learning_starts'

2

Hi! I am trying to use JSRL with stable baselines 3 PPO. However, at the line ``` model = jsrl.get_jsrl_algorithm(PPO)( "MultiInputPolicy", env, policy_kwargs=dict( guide_policy=guide_policy, max_horizon=max_horizon, strategy="curriculum", horizons=np.arange(max_horizon, -1, -max_horizon //...

aditink

Does the script of on-policy JSRL work? The commit message suggests it's not complete yet.

gaurav-gaurav

AttributeError: can't set attribute

1

I am trying to run the "examples/train_jsrl_curriculum.py", but there is error occured at the file "jsrl/jsrl.py" ` eval_callback = JSRLEvalCallback( self.env, callback_after_eval=JSRLAfterEvalCallback( self.policy, self.logger, verbose=self.verbose, ), eval_freq=self.policy.eval_freq, n_eval_episodes=self.policy.n_eval_episodes, verbose=self.verbose, )`...

astonisingHSR