5e63df5e0b53b583b6865cc099cecd5436563113,samples/reinforce.py,,,#,13

Before Change

    agent = ptan.agent.PolicyAgent(model)
    exp_source = ptan.experience.ExperienceSource(env=env, agent=agent, steps_count=run.getint("defaults", "n_steps"))

    for exp in exp_source:
        print(exp)

    pass

After Change

    exp_source = ptan.experience.ExperienceSource(env=env, agent=agent, steps_count=run.getint("defaults", "n_steps"))
    exp_buffer = ptan.experience.ExperienceReplayBuffer(exp_source, run.getint("exp_buffer", "size"))

    optimizer = optim.Adam(model.parameters(), lr=run.getfloat("learning", "lr"))

    def calc_loss(batch):
        
        Calculate loss expression from data batch

Instances