无法将保存的模型用作训练基准的MlpPolicy的起点?

时间:2018-09-11 17:04:24

标签: python tensorflow restore reinforcement-learning openai-gym

我目前正在train.py中使用以下代码,使用OpenAI基线中的代码来训练模型:

from baselines.common import tf_util as U
import tensorflow as tf
import gym, logging

from visak_dartdeepmimic import VisakDartDeepMimicArgParse

def train(env, initial_params_path,
        save_interval, out_prefix, num_timesteps, num_cpus):
    from baselines.ppo1 import mlp_policy, pposgd_simple
    sess = U.make_session(num_cpu=num_cpus).__enter__()

    U.initialize()

    def policy_fn(name, ob_space, ac_space):
        print("Policy with name: ", name)
        policy = mlp_policy.MlpPolicy(name=name, ob_space=ob_space, ac_space=ac_space,
            hid_size=64, num_hid_layers=2)
        saver = tf.train.Saver()
        if initial_params_path is not None:
            print("Tried to restore from ", initial_params_path)
            saver.restore(tf.get_default_session(), initial_params_path)
        return policy

    def callback_fn(local_vars, global_vars):
        iters = local_vars["iters_so_far"]
        saver = tf.train.Saver()
        if iters % save_interval == 0:
            saver.save(sess, out_prefix + str(iters))

    pposgd_simple.learn(env, policy_fn,
            max_timesteps=num_timesteps,
            callback=callback_fn,
            timesteps_per_actorbatch=2048,
            clip_param=0.2, entcoeff=0.0,
            optim_epochs=10, optim_stepsize=3e-4, optim_batchsize=64,
            gamma=1.0, lam=0.95, schedule='linear',
        )
    env.close()

这基于OpenAI本身提供的代码in the baselines repository

这很好用,除了我发现一些看起来很奇怪的学习曲线外,我怀疑这是由于传递给learn函数的一些超参数会导致性能随着事情的进行而衰减/高方差(尽管我不这样做)不确定)

enter image description here

无论如何,要确认这个假设,我想重新训练模型,而不是从头开始:我想从一个高点开始:例如,迭代1600,我有一个保存的模型在周围(具有在saver.save中用callback_fn保存了它

所以现在我调用train函数,但是这次我为它提供了一个指向迭代1600的保存前缀的inital_params_path。据我所知,对saver.restore的调用在policy_fn应该将模型恢复到“重置”到1teration 1600的位置(并且我已经确认加载例程是使用print语句运行的)

但是,实际上,我发现几乎没有任何东西被加载。例如,如果我得到

之类的统计信息
----------------------------------
| EpLenMean       | 74.2         |
| EpRewMean       | 38.7         |
| EpThisIter      | 209          |
| EpisodesSoFar   | 662438       |
| TimeElapsed     | 2.15e+04     |
| TimestepsSoFar  | 26230266     |
| ev_tdlam_before | 0.95         |
| loss_ent        | 2.7640965    |
| loss_kl         | 0.09064759   |
| loss_pol_entpen | 0.0          |
| loss_pol_surr   | -0.048767302 |
| loss_vf_loss    | 3.8620138    |
----------------------------------

对于迭代1600,然后对于新试验的迭代1(表面上使用1600的参数作为起点),我得到类似

----------------------------------
| EpLenMean       | 2.12         |
| EpRewMean       | 0.486        |
| EpThisIter      | 7676         |
| EpisodesSoFar   | 7676         |
| TimeElapsed     | 12.3         |
| TimestepsSoFar  | 16381        |
| ev_tdlam_before | -4.47        |
| loss_ent        | 45.355236    |
| loss_kl         | 0.016298374  |
| loss_pol_entpen | 0.0          |
| loss_pol_surr   | -0.039200217 |
| loss_vf_loss    | 0.043219414  |
----------------------------------

这又回到了平方(这是我的模型从头开始训练的地方)

有趣的是,我知道至少已正确保存了该模型,因为我实际上可以使用eval.py

重播该模型
from baselines.common import tf_util as U
from baselines.ppo1 import mlp_policy, pposgd_simple
import numpy as np
import tensorflow as tf

class PolicyLoaderAgent(object):
    """The world's simplest agent!"""
    def __init__(self, param_path, obs_space, action_space):
        self.action_space = action_space

        self.actor = mlp_policy.MlpPolicy("pi", obs_space, action_space,
                                        hid_size = 64, num_hid_layers=2)
        U.initialize()
        saver = tf.train.Saver()
        saver.restore(tf.get_default_session(), param_path)

    def act(self, observation, reward, done):
        action2, unknown = self.actor.act(False, observation)
        return action2


if __name__ == "__main__":

    parser = VisakDartDeepMimicArgParse()
    parser.add_argument("--params-prefix", required=True, type=str)
    args = parser.parse_args()
    env = parser.get_env()

    U.make_session(num_cpu=1).__enter__()

    U.initialize()

    agent = PolicyLoaderAgent(args.params_prefix, env.observation_space, env.action_space)

    while True:
        ob = env.reset(0, pos_stdv=0, vel_stdv=0)
        done = False
        while not done:
            action = agent.act(ob, reward, done)
            ob, reward, done, _ = env.step(action)
            env.render()

,我可以清楚地看到,与未经训练的基准相比,它学到了一些东西。两个文件的加载动作是相同的(或者,如果有一个错误,那么我找不到它),所以我发现train.py可能正确地加载了模型,然后由于某种原因在pposdg_simple.learn function's中,很快就忘记了。

有人可以阐明这种情况吗?

1 个答案:

答案 0 :(得分:0)

由于自发布此问题以来基线存储库已发生很大变化,因此不确定这是否仍然有意义,但是似乎您实际上并没有在恢复变量之前对其进行初始化。尝试将U.initialize()的呼叫移至policy_fn内:

def policy_fn(name, ob_space, ac_space):
    print("Policy with name: ", name)    
    policy = mlp_policy.MlpPolicy(name=name, ob_space=ob_space, 
                                  ac_space=ac_space, hid_size=64, num_hid_layers=2)
    saver = tf.train.Saver()
    if initial_params_path is not None:  
        print("Tried to restore from ", initial_params_path)
        U.initialize()
        saver.restore(tf.get_default_session(), initial_params_path)
    return policy