Python baselines.common.mpi_running_mean_std.RunningMeanStd() Examples

The following are 30 code examples of baselines.common.mpi_running_mean_std.RunningMeanStd(). You can vote up the ones you like or vote down the ones you don't like, and go to the original project or source file by following the links above each example. You may also want to check out all available functions/classes of the module baselines.common.mpi_running_mean_std , or try the search function .
Example #1
Source File: mlp_policy.py    From self-imitation-learning with MIT License 5 votes vote down vote up 
def _init(self, ob_space, ac_space, hid_size, num_hid_layers, gaussian_fixed_var=True):
        assert isinstance(ob_space, gym.spaces.Box)

        self.pdtype = pdtype = make_pdtype(ac_space)
        sequence_length = None

        ob = U.get_placeholder(name="ob", dtype=tf.float32, shape=[sequence_length] + list(ob_space.shape))

        with tf.variable_scope("obfilter"):
            self.ob_rms = RunningMeanStd(shape=ob_space.shape)

        obz = tf.clip_by_value((ob - self.ob_rms.mean) / self.ob_rms.std, -5.0, 5.0)
        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(dense(last_out, hid_size, "vffc%i" % (i+1), weight_init=U.normc_initializer(1.0)))
        self.vpred = dense(last_out, 1, "vffinal", weight_init=U.normc_initializer(1.0))[:, 0]

        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(dense(last_out, hid_size, "polfc%i" % (i+1), weight_init=U.normc_initializer(1.0)))

        if gaussian_fixed_var and isinstance(ac_space, gym.spaces.Box):
            mean = dense(last_out, pdtype.param_shape()[0]//2, "polfinal", U.normc_initializer(0.01))
            logstd = tf.get_variable(name="logstd", shape=[1, pdtype.param_shape()[0]//2], initializer=tf.zeros_initializer())
            pdparam = tf.concat([mean, mean * 0.0 + logstd], axis=1)
        else:
            pdparam = dense(last_out, pdtype.param_shape()[0], "polfinal", U.normc_initializer(0.01))

        self.pd = pdtype.pdfromflat(pdparam)

        self.state_in = []
        self.state_out = []

        # change for BC
        stochastic = U.get_placeholder(name="stochastic", dtype=tf.bool, shape=())
        ac = U.switch(stochastic, self.pd.sample(), self.pd.mode())
        self.ac = ac
        self._act = U.function([stochastic, ob], [ac, self.vpred])
Example #2
Source File: adversary.py    From lirpg with MIT License 5 votes vote down vote up 
def build_graph(self, obs_ph, acs_ph, reuse=False):
        with tf.variable_scope(self.scope):
            if reuse:
                tf.get_variable_scope().reuse_variables()

            with tf.variable_scope("obfilter"):
                self.obs_rms = RunningMeanStd(shape=self.observation_shape)
            obs = (obs_ph - self.obs_rms.mean / self.obs_rms.std)
            _input = tf.concat([obs, acs_ph], axis=1)  # concatenate the two input -> form a transition
            p_h1 = tf.contrib.layers.fully_connected(_input, self.hidden_size, activation_fn=tf.nn.tanh)
            p_h2 = tf.contrib.layers.fully_connected(p_h1, self.hidden_size, activation_fn=tf.nn.tanh)
            logits = tf.contrib.layers.fully_connected(p_h2, 1, activation_fn=tf.identity)
        return logits
Example #3
Source File: models.py    From ICML2019-TREX with MIT License 5 votes vote down vote up 
def _normalize_clip_observation(x, clip_range=[-5.0, 5.0]):
    rms = RunningMeanStd(shape=x.shape[1:])
    norm_x = tf.clip_by_value((x - rms.mean) / rms.std, min(clip_range), max(clip_range))
    return norm_x, rms
Example #4
Source File: mlp_policy.py    From ICML2019-TREX with MIT License 5 votes vote down vote up 
def _init(self, ob_space, ac_space, hid_size, num_hid_layers, gaussian_fixed_var=True):
        assert isinstance(ob_space, gym.spaces.Box)

        self.pdtype = pdtype = make_pdtype(ac_space)
        sequence_length = None

        ob = U.get_placeholder(name="ob", dtype=tf.float32, shape=[sequence_length] + list(ob_space.shape))

        with tf.variable_scope("obfilter"):
            self.ob_rms = RunningMeanStd(shape=ob_space.shape)

        obz = tf.clip_by_value((ob - self.ob_rms.mean) / self.ob_rms.std, -5.0, 5.0)
        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(dense(last_out, hid_size, "vffc%i" % (i+1), weight_init=U.normc_initializer(1.0)))
        self.vpred = dense(last_out, 1, "vffinal", weight_init=U.normc_initializer(1.0))[:, 0]

        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(dense(last_out, hid_size, "polfc%i" % (i+1), weight_init=U.normc_initializer(1.0)))

        if gaussian_fixed_var and isinstance(ac_space, gym.spaces.Box):
            mean = dense(last_out, pdtype.param_shape()[0]//2, "polfinal", U.normc_initializer(0.01))
            logstd = tf.get_variable(name="logstd", shape=[1, pdtype.param_shape()[0]//2], initializer=tf.zeros_initializer())
            pdparam = tf.concat([mean, mean * 0.0 + logstd], axis=1)
        else:
            pdparam = dense(last_out, pdtype.param_shape()[0], "polfinal", U.normc_initializer(0.01))

        self.pd = pdtype.pdfromflat(pdparam)

        self.state_in = []
        self.state_out = []

        # change for BC
        stochastic = U.get_placeholder(name="stochastic", dtype=tf.bool, shape=())
        ac = U.switch(stochastic, self.pd.sample(), self.pd.mode())
        self.ac = ac
        self._act = U.function([stochastic, ob], [ac, self.vpred])
Example #5
Source File: mlp_policy.py    From ICML2019-TREX with MIT License 5 votes vote down vote up 
def _init(self, ob_space, ac_space, hid_size, num_hid_layers, gaussian_fixed_var=True):
        assert isinstance(ob_space, gym.spaces.Box)

        self.pdtype = pdtype = make_pdtype(ac_space)
        sequence_length = None

        ob = U.get_placeholder(name="ob", dtype=tf.float32, shape=[sequence_length] + list(ob_space.shape))

        with tf.variable_scope("obfilter"):
            self.ob_rms = RunningMeanStd(shape=ob_space.shape)

        with tf.variable_scope('vf'):
            obz = tf.clip_by_value((ob - self.ob_rms.mean) / self.ob_rms.std, -5.0, 5.0)
            last_out = obz
            for i in range(num_hid_layers):
                last_out = tf.nn.tanh(tf.layers.dense(last_out, hid_size, name="fc%i"%(i+1), kernel_initializer=U.normc_initializer(1.0)))
            self.vpred = tf.layers.dense(last_out, 1, name='final', kernel_initializer=U.normc_initializer(1.0))[:,0]

        with tf.variable_scope('pol'):
            last_out = obz
            for i in range(num_hid_layers):
                last_out = tf.nn.tanh(tf.layers.dense(last_out, hid_size, name='fc%i'%(i+1), kernel_initializer=U.normc_initializer(1.0)))
            if gaussian_fixed_var and isinstance(ac_space, gym.spaces.Box):
                mean = tf.layers.dense(last_out, pdtype.param_shape()[0]//2, name='final', kernel_initializer=U.normc_initializer(0.01))
                logstd = tf.get_variable(name="logstd", shape=[1, pdtype.param_shape()[0]//2], initializer=tf.zeros_initializer())
                pdparam = tf.concat([mean, mean * 0.0 + logstd], axis=1)
            else:
                pdparam = tf.layers.dense(last_out, pdtype.param_shape()[0], name='final', kernel_initializer=U.normc_initializer(0.01))

        self.pd = pdtype.pdfromflat(pdparam)

        self.state_in = []
        self.state_out = []

        stochastic = tf.placeholder(dtype=tf.bool, shape=())
        ac = U.switch(stochastic, self.pd.sample(), self.pd.mode())
        self._act = U.function([stochastic, ob], [ac, self.vpred])
Example #6
Source File: policies.py    From ICML2019-TREX with MIT License 5 votes vote down vote up 
def _normalize_clip_observation(x, clip_range=[-5.0, 5.0]):
    rms = RunningMeanStd(shape=x.shape[1:])
    norm_x = tf.clip_by_value((x - rms.mean) / rms.std, min(clip_range), max(clip_range))
    return norm_x, rms
Example #7
Source File: models.py    From ICML2019-TREX with MIT License 5 votes vote down vote up 
def _normalize_clip_observation(x, clip_range=[-5.0, 5.0]):
    rms = RunningMeanStd(shape=x.shape[1:])
    norm_x = tf.clip_by_value((x - rms.mean) / rms.std, min(clip_range), max(clip_range))
    return norm_x, rms
Example #8
Source File: adversary.py    From sonic_contest with MIT License 5 votes vote down vote up 
def build_graph(self, obs_ph, acs_ph, reuse=False):
        with tf.variable_scope(self.scope):
            if reuse:
                tf.get_variable_scope().reuse_variables()

            with tf.variable_scope("obfilter"):
                self.obs_rms = RunningMeanStd(shape=self.observation_shape)
            obs = (obs_ph - self.obs_rms.mean / self.obs_rms.std)
            _input = tf.concat([obs, acs_ph], axis=1)  # concatenate the two input -> form a transition
            p_h1 = tf.contrib.layers.fully_connected(_input, self.hidden_size, activation_fn=tf.nn.tanh)
            p_h2 = tf.contrib.layers.fully_connected(p_h1, self.hidden_size, activation_fn=tf.nn.tanh)
            logits = tf.contrib.layers.fully_connected(p_h2, 1, activation_fn=tf.identity)
        return logits
Example #9
Source File: mlp_policy.py    From sonic_contest with MIT License 5 votes vote down vote up 
def _init(self, ob_space, ac_space, hid_size, num_hid_layers, gaussian_fixed_var=True):
        assert isinstance(ob_space, gym.spaces.Box)

        self.pdtype = pdtype = make_pdtype(ac_space)
        sequence_length = None

        ob = U.get_placeholder(name="ob", dtype=tf.float32, shape=[sequence_length] + list(ob_space.shape))

        with tf.variable_scope("obfilter"):
            self.ob_rms = RunningMeanStd(shape=ob_space.shape)

        obz = tf.clip_by_value((ob - self.ob_rms.mean) / self.ob_rms.std, -5.0, 5.0)
        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(dense(last_out, hid_size, "vffc%i" % (i+1), weight_init=U.normc_initializer(1.0)))
        self.vpred = dense(last_out, 1, "vffinal", weight_init=U.normc_initializer(1.0))[:, 0]

        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(dense(last_out, hid_size, "polfc%i" % (i+1), weight_init=U.normc_initializer(1.0)))

        if gaussian_fixed_var and isinstance(ac_space, gym.spaces.Box):
            mean = dense(last_out, pdtype.param_shape()[0]//2, "polfinal", U.normc_initializer(0.01))
            logstd = tf.get_variable(name="logstd", shape=[1, pdtype.param_shape()[0]//2], initializer=tf.zeros_initializer())
            pdparam = tf.concat([mean, mean * 0.0 + logstd], axis=1)
        else:
            pdparam = dense(last_out, pdtype.param_shape()[0], "polfinal", U.normc_initializer(0.01))

        self.pd = pdtype.pdfromflat(pdparam)

        self.state_in = []
        self.state_out = []

        # change for BC
        stochastic = U.get_placeholder(name="stochastic", dtype=tf.bool, shape=())
        ac = U.switch(stochastic, self.pd.sample(), self.pd.mode())
        self.ac = ac
        self._act = U.function([stochastic, ob], [ac, self.vpred])
Example #10
Source File: mlp_policy.py    From sonic_contest with MIT License 5 votes vote down vote up 
def _init(self, ob_space, ac_space, hid_size, num_hid_layers, gaussian_fixed_var=True):
        assert isinstance(ob_space, gym.spaces.Box)

        self.pdtype = pdtype = make_pdtype(ac_space)
        sequence_length = None

        ob = U.get_placeholder(name="ob", dtype=tf.float32, shape=[sequence_length] + list(ob_space.shape))

        with tf.variable_scope("obfilter"):
            self.ob_rms = RunningMeanStd(shape=ob_space.shape)

        with tf.variable_scope('vf'):
            obz = tf.clip_by_value((ob - self.ob_rms.mean) / self.ob_rms.std, -5.0, 5.0)
            last_out = obz
            for i in range(num_hid_layers):
                last_out = tf.nn.tanh(tf.layers.dense(last_out, hid_size, name="fc%i"%(i+1), kernel_initializer=U.normc_initializer(1.0)))
            self.vpred = tf.layers.dense(last_out, 1, name='final', kernel_initializer=U.normc_initializer(1.0))[:,0]

        with tf.variable_scope('pol'):
            last_out = obz
            for i in range(num_hid_layers):
                last_out = tf.nn.tanh(tf.layers.dense(last_out, hid_size, name='fc%i'%(i+1), kernel_initializer=U.normc_initializer(1.0)))
            if gaussian_fixed_var and isinstance(ac_space, gym.spaces.Box):
                mean = tf.layers.dense(last_out, pdtype.param_shape()[0]//2, name='final', kernel_initializer=U.normc_initializer(0.01))
                logstd = tf.get_variable(name="logstd", shape=[1, pdtype.param_shape()[0]//2], initializer=tf.zeros_initializer())
                pdparam = tf.concat([mean, mean * 0.0 + logstd], axis=1)
            else:
                pdparam = tf.layers.dense(last_out, pdtype.param_shape()[0], name='final', kernel_initializer=U.normc_initializer(0.01))

        self.pd = pdtype.pdfromflat(pdparam)

        self.state_in = []
        self.state_out = []

        stochastic = tf.placeholder(dtype=tf.bool, shape=())
        ac = U.switch(stochastic, self.pd.sample(), self.pd.mode())
        self._act = U.function([stochastic, ob], [ac, self.vpred])
Example #11
Source File: adversary.py    From self-imitation-learning with MIT License 5 votes vote down vote up 
def build_graph(self, obs_ph, acs_ph, reuse=False):
        with tf.variable_scope(self.scope):
            if reuse:
                tf.get_variable_scope().reuse_variables()

            with tf.variable_scope("obfilter"):
                self.obs_rms = RunningMeanStd(shape=self.observation_shape)
            obs = (obs_ph - self.obs_rms.mean / self.obs_rms.std)
            _input = tf.concat([obs, acs_ph], axis=1)  # concatenate the two input -> form a transition
            p_h1 = tf.contrib.layers.fully_connected(_input, self.hidden_size, activation_fn=tf.nn.tanh)
            p_h2 = tf.contrib.layers.fully_connected(p_h1, self.hidden_size, activation_fn=tf.nn.tanh)
            logits = tf.contrib.layers.fully_connected(p_h2, 1, activation_fn=tf.identity)
        return logits
Example #12
Source File: policies.py    From ICML2019-TREX with MIT License 5 votes vote down vote up 
def _normalize_clip_observation(x, clip_range=[-5.0, 5.0]):
    rms = RunningMeanStd(shape=x.shape[1:])
    norm_x = tf.clip_by_value((x - rms.mean) / rms.std, min(clip_range), max(clip_range))
    return norm_x, rms
Example #13
Source File: mlp_policy.py    From self-imitation-learning with MIT License 5 votes vote down vote up 
def _init(self, ob_space, ac_space, hid_size, num_hid_layers, gaussian_fixed_var=True):
        assert isinstance(ob_space, gym.spaces.Box)

        self.pdtype = pdtype = make_pdtype(ac_space)
        sequence_length = None

        ob = U.get_placeholder(name="ob", dtype=tf.float32, shape=[sequence_length] + list(ob_space.shape))

        with tf.variable_scope("obfilter"):
            self.ob_rms = RunningMeanStd(shape=ob_space.shape)

        with tf.variable_scope('vf'):
            obz = tf.clip_by_value((ob - self.ob_rms.mean) / self.ob_rms.std, -5.0, 5.0)
            last_out = obz
            for i in range(num_hid_layers):
                last_out = tf.nn.tanh(tf.layers.dense(last_out, hid_size, name="fc%i"%(i+1), kernel_initializer=U.normc_initializer(1.0)))
            self.vpred = tf.layers.dense(last_out, 1, name='final', kernel_initializer=U.normc_initializer(1.0))[:,0]

        with tf.variable_scope('pol'):
            last_out = obz
            for i in range(num_hid_layers):
                last_out = tf.nn.tanh(tf.layers.dense(last_out, hid_size, name='fc%i'%(i+1), kernel_initializer=U.normc_initializer(1.0)))
            if gaussian_fixed_var and isinstance(ac_space, gym.spaces.Box):
                mean = tf.layers.dense(last_out, pdtype.param_shape()[0]//2, name='final', kernel_initializer=U.normc_initializer(0.01))
                logstd = tf.get_variable(name="logstd", shape=[1, pdtype.param_shape()[0]//2], initializer=tf.zeros_initializer())
                pdparam = tf.concat([mean, mean * 0.0 + logstd], axis=1)
            else:
                pdparam = tf.layers.dense(last_out, pdtype.param_shape()[0], name='final', kernel_initializer=U.normc_initializer(0.01))

        self.pd = pdtype.pdfromflat(pdparam)

        self.state_in = []
        self.state_out = []

        stochastic = tf.placeholder(dtype=tf.bool, shape=())
        ac = U.switch(stochastic, self.pd.sample(), self.pd.mode())
        self._act = U.function([stochastic, ob], [ac, self.vpred])
Example #14
Source File: adversary.py    From baselines with MIT License 5 votes vote down vote up 
def build_graph(self, obs_ph, acs_ph, reuse=False):
        with tf.variable_scope(self.scope):
            if reuse:
                tf.get_variable_scope().reuse_variables()

            with tf.variable_scope("obfilter"):
                self.obs_rms = RunningMeanStd(shape=self.observation_shape)
            obs = (obs_ph - self.obs_rms.mean) / self.obs_rms.std
            _input = tf.concat([obs, acs_ph], axis=1)  # concatenate the two input -> form a transition
            p_h1 = tf.contrib.layers.fully_connected(_input, self.hidden_size, activation_fn=tf.nn.tanh)
            p_h2 = tf.contrib.layers.fully_connected(p_h1, self.hidden_size, activation_fn=tf.nn.tanh)
            logits = tf.contrib.layers.fully_connected(p_h2, 1, activation_fn=tf.identity)
        return logits
Example #15
Source File: mlp_policy.py    From baselines with MIT License 5 votes vote down vote up 
def _init(self, ob_space, ac_space, hid_size, num_hid_layers, gaussian_fixed_var=True):
        assert isinstance(ob_space, gym.spaces.Box)

        self.pdtype = pdtype = make_pdtype(ac_space)
        sequence_length = None

        ob = U.get_placeholder(name="ob", dtype=tf.float32, shape=[sequence_length] + list(ob_space.shape))

        with tf.variable_scope("obfilter"):
            self.ob_rms = RunningMeanStd(shape=ob_space.shape)

        obz = tf.clip_by_value((ob - self.ob_rms.mean) / self.ob_rms.std, -5.0, 5.0)
        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(dense(last_out, hid_size, "vffc%i" % (i+1), weight_init=U.normc_initializer(1.0)))
        self.vpred = dense(last_out, 1, "vffinal", weight_init=U.normc_initializer(1.0))[:, 0]

        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(dense(last_out, hid_size, "polfc%i" % (i+1), weight_init=U.normc_initializer(1.0)))

        if gaussian_fixed_var and isinstance(ac_space, gym.spaces.Box):
            mean = dense(last_out, pdtype.param_shape()[0]//2, "polfinal", U.normc_initializer(0.01))
            logstd = tf.get_variable(name="logstd", shape=[1, pdtype.param_shape()[0]//2], initializer=tf.zeros_initializer())
            pdparam = tf.concat([mean, mean * 0.0 + logstd], axis=1)
        else:
            pdparam = dense(last_out, pdtype.param_shape()[0], "polfinal", U.normc_initializer(0.01))

        self.pd = pdtype.pdfromflat(pdparam)

        self.state_in = []
        self.state_out = []

        # change for BC
        stochastic = U.get_placeholder(name="stochastic", dtype=tf.bool, shape=())
        ac = U.switch(stochastic, self.pd.sample(), self.pd.mode())
        self.ac = ac
        self._act = U.function([stochastic, ob], [ac, self.vpred])
Example #16
Source File: mlp_policy.py    From baselines with MIT License 5 votes vote down vote up 
def _init(self, ob_space, ac_space, hid_size, num_hid_layers, gaussian_fixed_var=True):
        assert isinstance(ob_space, gym.spaces.Box)

        self.pdtype = pdtype = make_pdtype(ac_space)
        sequence_length = None

        ob = U.get_placeholder(name="ob", dtype=tf.float32, shape=[sequence_length] + list(ob_space.shape))

        with tf.variable_scope("obfilter"):
            self.ob_rms = RunningMeanStd(shape=ob_space.shape)

        with tf.variable_scope('vf'):
            obz = tf.clip_by_value((ob - self.ob_rms.mean) / self.ob_rms.std, -5.0, 5.0)
            last_out = obz
            for i in range(num_hid_layers):
                last_out = tf.nn.tanh(tf.layers.dense(last_out, hid_size, name="fc%i"%(i+1), kernel_initializer=U.normc_initializer(1.0)))
            self.vpred = tf.layers.dense(last_out, 1, name='final', kernel_initializer=U.normc_initializer(1.0))[:,0]

        with tf.variable_scope('pol'):
            last_out = obz
            for i in range(num_hid_layers):
                last_out = tf.nn.tanh(tf.layers.dense(last_out, hid_size, name='fc%i'%(i+1), kernel_initializer=U.normc_initializer(1.0)))
            if gaussian_fixed_var and isinstance(ac_space, gym.spaces.Box):
                mean = tf.layers.dense(last_out, pdtype.param_shape()[0]//2, name='final', kernel_initializer=U.normc_initializer(0.01))
                logstd = tf.get_variable(name="logstd", shape=[1, pdtype.param_shape()[0]//2], initializer=tf.zeros_initializer())
                pdparam = tf.concat([mean, mean * 0.0 + logstd], axis=1)
            else:
                pdparam = tf.layers.dense(last_out, pdtype.param_shape()[0], name='final', kernel_initializer=U.normc_initializer(0.01))

        self.pd = pdtype.pdfromflat(pdparam)

        self.state_in = []
        self.state_out = []

        stochastic = tf.placeholder(dtype=tf.bool, shape=())
        ac = U.switch(stochastic, self.pd.sample(), self.pd.mode())
        self._act = U.function([stochastic, ob], [ac, self.vpred])
Example #17
Source File: policies.py    From baselines with MIT License 5 votes vote down vote up 
def _normalize_clip_observation(x, clip_range=[-5.0, 5.0]):
    rms = RunningMeanStd(shape=x.shape[1:])
    norm_x = tf.clip_by_value((x - rms.mean) / rms.std, min(clip_range), max(clip_range))
    return norm_x, rms
Example #18
Source File: models.py    From baselines with MIT License 5 votes vote down vote up 
def _normalize_clip_observation(x, clip_range=[-5.0, 5.0]):
    rms = RunningMeanStd(shape=x.shape[1:])
    norm_x = tf.clip_by_value((x - rms.mean) / rms.std, min(clip_range), max(clip_range))
    return norm_x, rms
Example #19
Source File: mlp_policy.py    From midlevel-reps with MIT License 5 votes vote down vote up 
def _init(self, ob_space, ac_space, hid_size, num_hid_layers, gaussian_fixed_var=True):
        
        assert isinstance(ob_space, gym.spaces.Box)

        self.pdtype = pdtype = make_pdtype(ac_space)
        sequence_length = None

        ob = U.get_placeholder(name="ob", dtype=tf.float32, shape=[sequence_length] + list(ob_space.shape))
        
        with tf.variable_scope("obfilter"):
            self.ob_rms = RunningMeanStd(shape=ob_space.shape)

        obz = tf.clip_by_value((ob - self.ob_rms.mean) / self.ob_rms.std, -5.0, 5.0)
        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(tf.layers.dense(last_out, hid_size, name="vffc%i"%(i+1), kernel_initializer=U.normc_initializer(1.0)))
        self.vpred = tf.layers.dense(last_out, 1, name="vffinal", kernel_initializer=U.normc_initializer(1.0))[:,0]
        
        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(tf.layers.dense(last_out, hid_size, name="polfc%i"%(i+1), kernel_initializer=U.normc_initializer(1.0)))
        if gaussian_fixed_var and isinstance(ac_space, gym.spaces.Box):
            mean = tf.layers.dense(last_out, pdtype.param_shape()[0]//2, name="polfinal", kernel_initializer=U.normc_initializer(0.01))
            logstd = tf.get_variable(name="logstd", shape=[1, pdtype.param_shape()[0]//2], initializer=tf.zeros_initializer())
            pdparam = tf.concat([mean, mean * 0.0 + logstd], axis=1)
        else:
            pdparam = tf.layers.dense(last_out, pdtype.param_shape()[0], name="polfinal", kernel_initializer=U.normc_initializer(0.01))

        self.pd = pdtype.pdfromflat(pdparam)

        self.state_in = []
        self.state_out = []

        stochastic = tf.placeholder(dtype=tf.bool, shape=())
        ac = U.switch(stochastic, self.pd.sample(), self.pd.mode())
        self._act = U.function([stochastic, ob], [ac, self.vpred])
Example #20
Source File: mlp_policy.py    From deeprl-baselines with MIT License 5 votes vote down vote up 
def _init(self, ob_space, ac_space, hid_size, num_hid_layers, gaussian_fixed_var=True):
        assert isinstance(ob_space, gym.spaces.Box)

        self.pdtype = pdtype = make_pdtype(ac_space)
        sequence_length = None

        ob = U.get_placeholder(name="ob", dtype=tf.float32, shape=[sequence_length] + list(ob_space.shape))
        
        with tf.variable_scope("obfilter"):
            self.ob_rms = RunningMeanStd(shape=ob_space.shape)

        obz = tf.clip_by_value((ob - self.ob_rms.mean) / self.ob_rms.std, -5.0, 5.0)
        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(U.dense(last_out, hid_size, "vffc%i"%(i+1), weight_init=U.normc_initializer(1.0)))
        self.vpred = U.dense(last_out, 1, "vffinal", weight_init=U.normc_initializer(1.0))[:,0]
        
        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(U.dense(last_out, hid_size, "polfc%i"%(i+1), weight_init=U.normc_initializer(1.0)))
        if gaussian_fixed_var and isinstance(ac_space, gym.spaces.Box):
            mean = U.dense(last_out, pdtype.param_shape()[0]//2, "polfinal", U.normc_initializer(0.01))            
            logstd = tf.get_variable(name="logstd", shape=[1, pdtype.param_shape()[0]//2], initializer=tf.zeros_initializer())
            pdparam = U.concatenate([mean, mean * 0.0 + logstd], axis=1)
        else:
            pdparam = U.dense(last_out, pdtype.param_shape()[0], "polfinal", U.normc_initializer(0.01))

        self.pd = pdtype.pdfromflat(pdparam)

        self.state_in = []
        self.state_out = []

        stochastic = tf.placeholder(dtype=tf.bool, shape=())
        ac = U.switch(stochastic, self.pd.sample(), self.pd.mode())
        self._act = U.function([stochastic, ob], [ac, self.vpred])
Example #21
Source File: mlp_policy.py    From BackpropThroughTheVoidRL with MIT License 5 votes vote down vote up 
def _init(self, ob_space, ac_space, hid_size, num_hid_layers, gaussian_fixed_var=True):
        assert isinstance(ob_space, gym.spaces.Box)

        self.pdtype = pdtype = make_pdtype(ac_space)
        sequence_length = None

        ob = U.get_placeholder(name="ob", dtype=tf.float32, shape=[sequence_length] + list(ob_space.shape))
        
        with tf.variable_scope("obfilter"):
            self.ob_rms = RunningMeanStd(shape=ob_space.shape)

        obz = tf.clip_by_value((ob - self.ob_rms.mean) / self.ob_rms.std, -5.0, 5.0)
        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(U.dense(last_out, hid_size, "vffc%i"%(i+1), weight_init=U.normc_initializer(1.0)))
        self.vpred = U.dense(last_out, 1, "vffinal", weight_init=U.normc_initializer(1.0))[:,0]
        
        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(U.dense(last_out, hid_size, "polfc%i"%(i+1), weight_init=U.normc_initializer(1.0)))
        if gaussian_fixed_var and isinstance(ac_space, gym.spaces.Box):
            mean = U.dense(last_out, pdtype.param_shape()[0]//2, "polfinal", U.normc_initializer(0.01))            
            logstd = tf.get_variable(name="logstd", shape=[1, pdtype.param_shape()[0]//2], initializer=tf.zeros_initializer())
            pdparam = U.concatenate([mean, mean * 0.0 + logstd], axis=1)
        else:
            pdparam = U.dense(last_out, pdtype.param_shape()[0], "polfinal", U.normc_initializer(0.01))

        self.pd = pdtype.pdfromflat(pdparam)

        self.state_in = []
        self.state_out = []

        stochastic = tf.placeholder(dtype=tf.bool, shape=())
        ac = U.switch(stochastic, self.pd.sample(), self.pd.mode())
        self._act = U.function([stochastic, ob], [ac, self.vpred])
Example #22
Source File: policies.py    From Reinforcement_Learning_for_Traffic_Light_Control with Apache License 2.0 5 votes vote down vote up 
def _normalize_clip_observation(x, clip_range=[-5.0, 5.0]):
    rms = RunningMeanStd(shape=x.shape[1:])
    norm_x = tf.clip_by_value((x - rms.mean) / rms.std, min(clip_range), max(clip_range))
    return norm_x, rms
Example #23
Source File: mlp_policy.py    From lirpg with MIT License 5 votes vote down vote up 
def _init(self, ob_space, ac_space, hid_size, num_hid_layers, gaussian_fixed_var=True):
        assert isinstance(ob_space, gym.spaces.Box)

        self.pdtype = pdtype = make_pdtype(ac_space)
        sequence_length = None

        ob = U.get_placeholder(name="ob", dtype=tf.float32, shape=[sequence_length] + list(ob_space.shape))

        with tf.variable_scope("obfilter"):
            self.ob_rms = RunningMeanStd(shape=ob_space.shape)

        obz = tf.clip_by_value((ob - self.ob_rms.mean) / self.ob_rms.std, -5.0, 5.0)
        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(dense(last_out, hid_size, "vffc%i" % (i+1), weight_init=U.normc_initializer(1.0)))
        self.vpred = dense(last_out, 1, "vffinal", weight_init=U.normc_initializer(1.0))[:, 0]

        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(dense(last_out, hid_size, "polfc%i" % (i+1), weight_init=U.normc_initializer(1.0)))

        if gaussian_fixed_var and isinstance(ac_space, gym.spaces.Box):
            mean = dense(last_out, pdtype.param_shape()[0]//2, "polfinal", U.normc_initializer(0.01))
            logstd = tf.get_variable(name="logstd", shape=[1, pdtype.param_shape()[0]//2], initializer=tf.zeros_initializer())
            pdparam = tf.concat([mean, mean * 0.0 + logstd], axis=1)
        else:
            pdparam = dense(last_out, pdtype.param_shape()[0], "polfinal", U.normc_initializer(0.01))

        self.pd = pdtype.pdfromflat(pdparam)

        self.state_in = []
        self.state_out = []

        # change for BC
        stochastic = U.get_placeholder(name="stochastic", dtype=tf.bool, shape=())
        ac = U.switch(stochastic, self.pd.sample(), self.pd.mode())
        self.ac = ac
        self._act = U.function([stochastic, ob], [ac, self.vpred])
Example #24
Source File: mlp_policy.py    From lirpg with MIT License 5 votes vote down vote up 
def _init(self, ob_space, ac_space, hid_size, num_hid_layers, gaussian_fixed_var=True):
        assert isinstance(ob_space, gym.spaces.Box)

        self.pdtype = pdtype = make_pdtype(ac_space)
        sequence_length = None

        ob = U.get_placeholder(name="ob", dtype=tf.float32, shape=[sequence_length] + list(ob_space.shape))

        with tf.variable_scope("obfilter"):
            self.ob_rms = RunningMeanStd(shape=ob_space.shape)

        with tf.variable_scope('vf'):
            obz = tf.clip_by_value((ob - self.ob_rms.mean) / self.ob_rms.std, -5.0, 5.0)
            last_out = obz
            for i in range(num_hid_layers):
                last_out = tf.nn.tanh(tf.layers.dense(last_out, hid_size, name="fc%i"%(i+1), kernel_initializer=U.normc_initializer(1.0)))
            self.vpred = tf.layers.dense(last_out, 1, name='final', kernel_initializer=U.normc_initializer(1.0))[:,0]

        with tf.variable_scope('pol'):
            last_out = obz
            for i in range(num_hid_layers):
                last_out = tf.nn.tanh(tf.layers.dense(last_out, hid_size, name='fc%i'%(i+1), kernel_initializer=U.normc_initializer(1.0)))
            if gaussian_fixed_var and isinstance(ac_space, gym.spaces.Box):
                mean = tf.layers.dense(last_out, pdtype.param_shape()[0]//2, name='final', kernel_initializer=U.normc_initializer(0.01))
                logstd = tf.get_variable(name="logstd", shape=[1, pdtype.param_shape()[0]//2], initializer=tf.zeros_initializer())
                pdparam = tf.concat([mean, mean * 0.0 + logstd], axis=1)
            else:
                pdparam = tf.layers.dense(last_out, pdtype.param_shape()[0], name='final', kernel_initializer=U.normc_initializer(0.01))

        self.pd = pdtype.pdfromflat(pdparam)

        self.state_in = []
        self.state_out = []

        stochastic = tf.placeholder(dtype=tf.bool, shape=())
        ac = U.switch(stochastic, self.pd.sample(), self.pd.mode())
        self._act = U.function([stochastic, ob], [ac, self.vpred])
Example #25
Source File: adversary.py    From HardRLWithYoutube with MIT License 5 votes vote down vote up 
def build_graph(self, obs_ph, acs_ph, reuse=False):
        with tf.variable_scope(self.scope):
            if reuse:
                tf.get_variable_scope().reuse_variables()

            with tf.variable_scope("obfilter"):
                self.obs_rms = RunningMeanStd(shape=self.observation_shape)
            obs = (obs_ph - self.obs_rms.mean / self.obs_rms.std)
            _input = tf.concat([obs, acs_ph], axis=1)  # concatenate the two input -> form a transition
            p_h1 = tf.contrib.layers.fully_connected(_input, self.hidden_size, activation_fn=tf.nn.tanh)
            p_h2 = tf.contrib.layers.fully_connected(p_h1, self.hidden_size, activation_fn=tf.nn.tanh)
            logits = tf.contrib.layers.fully_connected(p_h2, 1, activation_fn=tf.identity)
        return logits
Example #26
Source File: mlp_policy.py    From HardRLWithYoutube with MIT License 5 votes vote down vote up 
def _init(self, ob_space, ac_space, hid_size, num_hid_layers, gaussian_fixed_var=True):
        assert isinstance(ob_space, gym.spaces.Box)

        self.pdtype = pdtype = make_pdtype(ac_space)
        sequence_length = None

        ob = U.get_placeholder(name="ob", dtype=tf.float32, shape=[sequence_length] + list(ob_space.shape))

        with tf.variable_scope("obfilter"):
            self.ob_rms = RunningMeanStd(shape=ob_space.shape)

        obz = tf.clip_by_value((ob - self.ob_rms.mean) / self.ob_rms.std, -5.0, 5.0)
        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(dense(last_out, hid_size, "vffc%i" % (i+1), weight_init=U.normc_initializer(1.0)))
        self.vpred = dense(last_out, 1, "vffinal", weight_init=U.normc_initializer(1.0))[:, 0]

        last_out = obz
        for i in range(num_hid_layers):
            last_out = tf.nn.tanh(dense(last_out, hid_size, "polfc%i" % (i+1), weight_init=U.normc_initializer(1.0)))

        if gaussian_fixed_var and isinstance(ac_space, gym.spaces.Box):
            mean = dense(last_out, pdtype.param_shape()[0]//2, "polfinal", U.normc_initializer(0.01))
            logstd = tf.get_variable(name="logstd", shape=[1, pdtype.param_shape()[0]//2], initializer=tf.zeros_initializer())
            pdparam = tf.concat([mean, mean * 0.0 + logstd], axis=1)
        else:
            pdparam = dense(last_out, pdtype.param_shape()[0], "polfinal", U.normc_initializer(0.01))

        self.pd = pdtype.pdfromflat(pdparam)

        self.state_in = []
        self.state_out = []

        # change for BC
        stochastic = U.get_placeholder(name="stochastic", dtype=tf.bool, shape=())
        ac = U.switch(stochastic, self.pd.sample(), self.pd.mode())
        self.ac = ac
        self._act = U.function([stochastic, ob], [ac, self.vpred])
Example #27
Source File: mlp_policy.py    From HardRLWithYoutube with MIT License 5 votes vote down vote up 
def _init(self, ob_space, ac_space, hid_size, num_hid_layers, gaussian_fixed_var=True):
        assert isinstance(ob_space, gym.spaces.Box)

        self.pdtype = pdtype = make_pdtype(ac_space)
        sequence_length = None

        ob = U.get_placeholder(name="ob", dtype=tf.float32, shape=[sequence_length] + list(ob_space.shape))

        with tf.variable_scope("obfilter"):
            self.ob_rms = RunningMeanStd(shape=ob_space.shape)

        with tf.variable_scope('vf'):
            obz = tf.clip_by_value((ob - self.ob_rms.mean) / self.ob_rms.std, -5.0, 5.0)
            last_out = obz
            for i in range(num_hid_layers):
                last_out = tf.nn.tanh(tf.layers.dense(last_out, hid_size, name="fc%i"%(i+1), kernel_initializer=U.normc_initializer(1.0)))
            self.vpred = tf.layers.dense(last_out, 1, name='final', kernel_initializer=U.normc_initializer(1.0))[:,0]

        with tf.variable_scope('pol'):
            last_out = obz
            for i in range(num_hid_layers):
                last_out = tf.nn.tanh(tf.layers.dense(last_out, hid_size, name='fc%i'%(i+1), kernel_initializer=U.normc_initializer(1.0)))
            if gaussian_fixed_var and isinstance(ac_space, gym.spaces.Box):
                mean = tf.layers.dense(last_out, pdtype.param_shape()[0]//2, name='final', kernel_initializer=U.normc_initializer(0.01))
                logstd = tf.get_variable(name="logstd", shape=[1, pdtype.param_shape()[0]//2], initializer=tf.zeros_initializer())
                pdparam = tf.concat([mean, mean * 0.0 + logstd], axis=1)
            else:
                pdparam = tf.layers.dense(last_out, pdtype.param_shape()[0], name='final', kernel_initializer=U.normc_initializer(0.01))

        self.pd = pdtype.pdfromflat(pdparam)

        self.state_in = []
        self.state_out = []

        stochastic = tf.placeholder(dtype=tf.bool, shape=())
        ac = U.switch(stochastic, self.pd.sample(), self.pd.mode())
        self._act = U.function([stochastic, ob], [ac, self.vpred])
Example #28
Source File: policies.py    From HardRLWithYoutube with MIT License 5 votes vote down vote up 
def _normalize_clip_observation(x, clip_range=[-5.0, 5.0]):
    rms = RunningMeanStd(shape=x.shape[1:])
    norm_x = tf.clip_by_value((x - rms.mean) / rms.std, min(clip_range), max(clip_range))
    return norm_x, rms
Example #29
Source File: models.py    From HardRLWithYoutube with MIT License 5 votes vote down vote up 
def _normalize_clip_observation(x, clip_range=[-5.0, 5.0]):
    rms = RunningMeanStd(shape=x.shape[1:])
    norm_x = tf.clip_by_value((x - rms.mean) / rms.std, min(clip_range), max(clip_range))
    return norm_x, rms
Example #30
Source File: policies.py    From Reinforcement_Learning_for_Traffic_Light_Control with Apache License 2.0 5 votes vote down vote up 
def _normalize_clip_observation(x, clip_range=[-5.0, 5.0]):
    rms = RunningMeanStd(shape=x.shape[1:])
    norm_x = tf.clip_by_value((x - rms.mean) / rms.std, min(clip_range), max(clip_range))
    return norm_x, rms