2023-03-31 15:59:42 +00:00
|
|
|
import numpy as np
|
|
|
|
|
2023-03-23 14:05:28 +00:00
|
|
|
import torch
|
|
|
|
import torch.nn as nn
|
|
|
|
import torch.nn.functional as F
|
|
|
|
from torch.distributions.normal import Normal
|
|
|
|
|
|
|
|
|
|
|
|
class ObservationEncoder(nn.Module):
|
|
|
|
def __init__(self, obs_shape, state_size, num_layers=4, num_filters=32, stride=None):
|
|
|
|
super().__init__()
|
|
|
|
|
|
|
|
assert len(obs_shape) == 3
|
|
|
|
|
|
|
|
self.state_size = state_size
|
|
|
|
|
|
|
|
layers = []
|
|
|
|
for i in range(num_layers):
|
|
|
|
input_channels = obs_shape[0] if i == 0 else output_channels
|
|
|
|
output_channels = num_filters * (2 ** i)
|
|
|
|
layers.append(nn.Conv2d(in_channels=input_channels, out_channels= output_channels, kernel_size=4, stride=2))
|
2023-04-10 18:18:39 +00:00
|
|
|
layers.append(nn.LeakyReLU())
|
2023-03-23 14:05:28 +00:00
|
|
|
|
|
|
|
self.convs = nn.Sequential(*layers)
|
|
|
|
|
|
|
|
self.fc = nn.Linear(256 * 3 * 3, 2 * state_size)
|
|
|
|
|
|
|
|
def forward(self, x):
|
|
|
|
x = self.convs(x)
|
|
|
|
x = x.view(x.size(0), -1)
|
|
|
|
x = self.fc(x)
|
|
|
|
|
|
|
|
# Mean and standard deviation
|
|
|
|
mean, std = torch.chunk(x, 2, dim=-1)
|
|
|
|
std = F.softplus(std)
|
|
|
|
std = torch.clamp(std, min=0.0, max=1e5)
|
|
|
|
|
|
|
|
# Normal Distribution
|
2023-03-27 17:22:17 +00:00
|
|
|
dist = self.get_dist(mean, std)
|
|
|
|
|
|
|
|
# Sampling via reparameterization Trick
|
2023-04-12 07:34:11 +00:00
|
|
|
#x = dist.rsample()
|
2023-03-23 14:05:28 +00:00
|
|
|
x = self.reparameterize(mean, std)
|
2023-03-27 17:22:17 +00:00
|
|
|
|
|
|
|
encoded_output = {"sample": x, "distribution": dist}
|
|
|
|
return encoded_output
|
2023-03-23 14:05:28 +00:00
|
|
|
|
|
|
|
def reparameterize(self, mu, std):
|
|
|
|
eps = torch.randn_like(std)
|
|
|
|
return mu + eps * std
|
2023-03-27 17:22:17 +00:00
|
|
|
|
|
|
|
def get_dist(self, mean, std):
|
|
|
|
distribution = torch.distributions.Normal(mean, std)
|
|
|
|
distribution = torch.distributions.independent.Independent(distribution, 1)
|
|
|
|
return distribution
|
2023-03-23 14:05:28 +00:00
|
|
|
|
|
|
|
|
|
|
|
class ObservationDecoder(nn.Module):
|
|
|
|
def __init__(self, state_size, output_shape):
|
|
|
|
super().__init__()
|
|
|
|
|
|
|
|
self.state_size = state_size
|
|
|
|
self.output_shape = output_shape
|
|
|
|
self.input_size = 256 * 3 * 3
|
|
|
|
self.in_channels = [self.input_size, 256, 128, 64]
|
|
|
|
self.out_channels = [256, 128, 64, 3]
|
|
|
|
|
|
|
|
if output_shape[1] == 84:
|
|
|
|
self.kernels = [5, 7, 5, 6]
|
|
|
|
self.output_padding = [1, 1, 1, 0]
|
|
|
|
elif output_shape[1] == 64:
|
|
|
|
self.kernels = [5, 5, 6, 6]
|
|
|
|
self.output_padding = [0, 0, 0, 0]
|
|
|
|
|
|
|
|
self.dense = nn.Linear(state_size, self.input_size)
|
|
|
|
|
|
|
|
layers = []
|
|
|
|
for i in range(len(self.kernels)):
|
|
|
|
layers.append(nn.ConvTranspose2d(in_channels=self.in_channels[i], out_channels=self.out_channels[i],
|
|
|
|
kernel_size=self.kernels[i], stride=2, output_padding=self.output_padding[i]))
|
|
|
|
if i!=len(self.kernels)-1:
|
2023-04-12 16:22:02 +00:00
|
|
|
layers.append(nn.LeakyReLU())
|
2023-03-23 14:05:28 +00:00
|
|
|
|
|
|
|
self.convtranspose = nn.Sequential(*layers)
|
|
|
|
|
|
|
|
def forward(self, features):
|
|
|
|
out_batch_shape = features.shape[:-1]
|
|
|
|
out = self.dense(features)
|
|
|
|
out = torch.reshape(out, [-1, self.input_size, 1, 1])
|
|
|
|
out = self.convtranspose(out)
|
|
|
|
mean = torch.reshape(out, (*out_batch_shape, *self.output_shape))
|
|
|
|
out_dist = torch.distributions.independent.Independent(torch.distributions.Normal(mean, 1), len(self.output_shape))
|
|
|
|
return out_dist
|
|
|
|
|
|
|
|
|
2023-04-10 11:18:41 +00:00
|
|
|
class Actor(nn.Module):
|
2023-03-31 15:59:42 +00:00
|
|
|
def __init__(self, state_size, hidden_size, action_size, num_layers=5):
|
|
|
|
super().__init__()
|
|
|
|
self.state_size = state_size
|
|
|
|
self.hidden_size = hidden_size
|
|
|
|
self.action_size = action_size
|
|
|
|
self.num_layers = num_layers
|
|
|
|
|
|
|
|
self._min_std=torch.Tensor([1e-4])[0]
|
|
|
|
self._init_std=torch.Tensor([5])[0]
|
|
|
|
self._mean_scale=torch.Tensor([5])[0]
|
|
|
|
|
|
|
|
layers = []
|
|
|
|
for i in range(self.num_layers):
|
|
|
|
input_channels = state_size if i == 0 else self.hidden_size
|
|
|
|
output_channels = self.hidden_size if i!= self.num_layers-1 else 2*action_size
|
|
|
|
layers.append(nn.Linear(input_channels, output_channels))
|
2023-04-12 16:22:02 +00:00
|
|
|
layers.append(nn.LeakyReLU())
|
2023-03-31 15:59:42 +00:00
|
|
|
self.action_model = nn.Sequential(*layers)
|
|
|
|
|
|
|
|
def get_dist(self, mean, std):
|
|
|
|
distribution = torch.distributions.Normal(mean, std)
|
|
|
|
distribution = torch.distributions.transformed_distribution.TransformedDistribution(distribution, TanhBijector())
|
|
|
|
distribution = torch.distributions.independent.Independent(distribution, 1)
|
|
|
|
return distribution
|
|
|
|
|
|
|
|
def forward(self, features):
|
|
|
|
out = self.action_model(features)
|
|
|
|
mean, std = torch.chunk(out, 2, dim=-1)
|
|
|
|
|
|
|
|
raw_init_std = torch.log(torch.exp(self._init_std) - 1)
|
|
|
|
action_mean = self._mean_scale * torch.tanh(mean / self._mean_scale)
|
|
|
|
action_std = F.softplus(std + raw_init_std) + self._min_std
|
|
|
|
|
|
|
|
dist = self.get_dist(action_mean, action_std)
|
|
|
|
sample = dist.rsample()
|
|
|
|
return sample
|
|
|
|
|
|
|
|
|
2023-03-31 16:38:51 +00:00
|
|
|
class ValueModel(nn.Module):
|
|
|
|
def __init__(self, state_size, hidden_size, num_layers=4):
|
|
|
|
super().__init__()
|
|
|
|
self.state_size = state_size
|
|
|
|
self.hidden_size = hidden_size
|
|
|
|
self.num_layers = num_layers
|
|
|
|
|
|
|
|
layers = []
|
|
|
|
for i in range(self.num_layers):
|
|
|
|
input_channels = state_size if i == 0 else self.hidden_size
|
|
|
|
output_channels = self.hidden_size if i!= self.num_layers-1 else 1
|
|
|
|
layers.append(nn.Linear(input_channels, output_channels))
|
2023-04-12 16:22:02 +00:00
|
|
|
layers.append(nn.LeakyReLU())
|
2023-03-31 16:38:51 +00:00
|
|
|
self.value_model = nn.Sequential(*layers)
|
|
|
|
|
|
|
|
def forward(self, state):
|
|
|
|
value = self.value_model(state)
|
2023-03-31 17:12:46 +00:00
|
|
|
value_dist = torch.distributions.independent.Independent(torch.distributions.Normal(value, 1), 1)
|
|
|
|
return value_dist
|
2023-04-10 11:18:41 +00:00
|
|
|
|
|
|
|
|
|
|
|
class RewardModel(nn.Module):
|
|
|
|
def __init__(self, state_size, hidden_size):
|
|
|
|
super().__init__()
|
|
|
|
self.reward_model = nn.Sequential(
|
|
|
|
nn.Linear(state_size, hidden_size),
|
2023-04-12 16:22:02 +00:00
|
|
|
nn.LeakyReLU(),
|
2023-04-10 11:18:41 +00:00
|
|
|
nn.Linear(hidden_size, hidden_size),
|
2023-04-12 16:22:02 +00:00
|
|
|
nn.LeakyReLU(),
|
2023-04-10 11:18:41 +00:00
|
|
|
nn.Linear(hidden_size, 1)
|
|
|
|
)
|
|
|
|
|
|
|
|
def forward(self, state):
|
2023-04-12 07:34:11 +00:00
|
|
|
reward = self.reward_model(state)
|
|
|
|
return torch.distributions.independent.Independent(
|
|
|
|
torch.distributions.Normal(reward, 1), 1)
|
2023-04-10 11:18:41 +00:00
|
|
|
|
2023-03-23 14:05:28 +00:00
|
|
|
class TransitionModel(nn.Module):
|
|
|
|
def __init__(self, state_size, hidden_size, action_size, history_size):
|
|
|
|
super().__init__()
|
|
|
|
|
|
|
|
self.state_size = state_size
|
|
|
|
self.hidden_size = hidden_size
|
|
|
|
self.action_size = action_size
|
|
|
|
self.history_size = history_size
|
2023-04-12 16:22:02 +00:00
|
|
|
self.act_fn = nn.LeakyReLU()
|
2023-03-23 14:05:28 +00:00
|
|
|
|
|
|
|
self.fc_state_action = nn.Linear(state_size + action_size, hidden_size)
|
|
|
|
self.history_cell = nn.GRUCell(hidden_size + history_size, history_size)
|
|
|
|
self.fc_state_prior = nn.Linear(history_size + state_size + action_size, 2 * state_size)
|
|
|
|
self.fc_state_posterior = nn.Linear(history_size + state_size + action_size, 2 * state_size)
|
|
|
|
|
|
|
|
def init_states(self, batch_size, device):
|
|
|
|
self.prev_state = torch.zeros(batch_size, self.state_size).to(device)
|
|
|
|
self.prev_action = torch.zeros(batch_size, self.action_size).to(device)
|
|
|
|
self.prev_history = torch.zeros(batch_size, self.history_size).to(device)
|
|
|
|
|
|
|
|
def get_dist(self, mean, std):
|
|
|
|
distribution = torch.distributions.Normal(mean, std)
|
|
|
|
distribution = torch.distributions.independent.Independent(distribution, 1)
|
|
|
|
return distribution
|
|
|
|
|
|
|
|
def imagine_step(self, prev_state, prev_action, prev_history):
|
|
|
|
state_action = self.act_fn(self.fc_state_action(torch.cat([prev_state, prev_action], dim=-1)))
|
2023-04-10 18:18:39 +00:00
|
|
|
prev_hist = prev_history.detach()
|
|
|
|
history = self.history_cell(torch.cat([state_action, prev_hist], dim=-1), prev_hist)
|
2023-03-23 14:05:28 +00:00
|
|
|
|
|
|
|
state_prior = self.fc_state_prior(torch.cat([history, prev_state, prev_action], dim=-1))
|
|
|
|
state_prior_mean, state_prior_std = torch.chunk(state_prior, 2, dim=-1)
|
|
|
|
state_prior_std = F.softplus(state_prior_std)
|
|
|
|
|
2023-03-27 17:22:17 +00:00
|
|
|
# Normal Distribution
|
|
|
|
state_prior_dist = self.get_dist(state_prior_mean, state_prior_std)
|
|
|
|
|
|
|
|
# Sampling via reparameterization Trick
|
2023-03-23 14:05:28 +00:00
|
|
|
sample_state_prior = self.reparemeterize(state_prior_mean, state_prior_std)
|
2023-03-27 17:22:17 +00:00
|
|
|
prior = {"mean": state_prior_mean, "std": state_prior_std, "sample": sample_state_prior, "history": history, "distribution": state_prior_dist}
|
2023-03-23 14:05:28 +00:00
|
|
|
return prior
|
|
|
|
|
2023-04-10 11:18:41 +00:00
|
|
|
def stack_states(self, states, dim=0):
|
2023-04-09 16:23:16 +00:00
|
|
|
s = dict(
|
|
|
|
mean = torch.stack([state['mean'] for state in states], dim=dim),
|
|
|
|
std = torch.stack([state['std'] for state in states], dim=dim),
|
|
|
|
sample = torch.stack([state['sample'] for state in states], dim=dim),
|
|
|
|
history = torch.stack([state['history'] for state in states], dim=dim),)
|
|
|
|
dist = dict(distribution = [state['distribution'] for state in states])
|
|
|
|
s.update(dist)
|
|
|
|
return s
|
|
|
|
|
|
|
|
def imagine_rollout(self, state, action, history, horizon):
|
|
|
|
imagined_priors = []
|
|
|
|
for i in range(horizon):
|
|
|
|
prior = self.imagine_step(state, action, history)
|
|
|
|
state = prior["sample"]
|
|
|
|
history = prior["history"]
|
|
|
|
imagined_priors.append(prior)
|
|
|
|
imagined_priors = self.stack_states(imagined_priors, dim=0)
|
|
|
|
return imagined_priors
|
|
|
|
|
2023-03-23 14:05:28 +00:00
|
|
|
def reparemeterize(self, mean, std):
|
|
|
|
eps = torch.randn_like(std)
|
|
|
|
return mean + eps * std
|
2023-03-31 15:59:42 +00:00
|
|
|
|
|
|
|
|
|
|
|
class TanhBijector(torch.distributions.Transform):
|
|
|
|
def __init__(self):
|
|
|
|
super().__init__()
|
|
|
|
self.bijective = True
|
|
|
|
self.domain = torch.distributions.constraints.real
|
|
|
|
self.codomain = torch.distributions.constraints.interval(-1.0, 1.0)
|
|
|
|
|
|
|
|
@property
|
|
|
|
def sign(self): return 1.
|
|
|
|
|
|
|
|
def _call(self, x): return torch.tanh(x)
|
|
|
|
|
|
|
|
def atanh(self, x):
|
|
|
|
return 0.5 * torch.log((1 + x) / (1 - x))
|
|
|
|
|
|
|
|
def _inverse(self, y: torch.Tensor):
|
|
|
|
y = torch.where(
|
|
|
|
(torch.abs(y) <= 1.),
|
|
|
|
torch.clamp(y, -0.99999997, 0.99999997),
|
|
|
|
y)
|
|
|
|
y = self.atanh(y)
|
|
|
|
return y
|
|
|
|
|
|
|
|
def log_abs_det_jacobian(self, x, y):
|
|
|
|
#return 2. * (np.log(2) - x - F.softplus(-2. * x))
|
|
|
|
return 2.0 * (torch.log(torch.tensor([2.0])) - x - F.softplus(-2.0 * x))
|
|
|
|
|
2023-03-23 14:05:28 +00:00
|
|
|
|
2023-04-02 16:52:26 +00:00
|
|
|
class ProjectionHead(nn.Module):
|
|
|
|
def __init__(self, state_size, action_size, hidden_size):
|
|
|
|
super(ProjectionHead, self).__init__()
|
|
|
|
self.state_size = state_size
|
|
|
|
self.action_size = action_size
|
|
|
|
self.hidden_size = hidden_size
|
|
|
|
|
|
|
|
self.projection_model = nn.Sequential(
|
|
|
|
nn.Linear(state_size + action_size, hidden_size),
|
|
|
|
nn.LayerNorm(hidden_size),
|
2023-04-12 16:22:02 +00:00
|
|
|
nn.LeakyReLU(),
|
2023-04-02 16:52:26 +00:00
|
|
|
nn.Linear(hidden_size, hidden_size),
|
|
|
|
nn.LayerNorm(hidden_size),
|
|
|
|
)
|
|
|
|
|
|
|
|
def forward(self, state, action):
|
|
|
|
x = torch.cat([state, action], dim=-1)
|
|
|
|
x = self.projection_model(x)
|
|
|
|
return x
|
|
|
|
|
|
|
|
|
|
|
|
class ContrastiveHead(nn.Module):
|
|
|
|
def __init__(self, hidden_size, temperature=1):
|
|
|
|
super(ContrastiveHead, self).__init__()
|
|
|
|
self.hidden_size = hidden_size
|
|
|
|
self.temperature = temperature
|
|
|
|
self.W = nn.Parameter(torch.rand(self.hidden_size, self.hidden_size))
|
|
|
|
|
|
|
|
def forward(self, z_a, z_pos):
|
|
|
|
Wz = torch.matmul(self.W, z_pos.T) # (z_dim,B)
|
|
|
|
logits = torch.matmul(z_a, Wz) # (B,B)
|
|
|
|
logits = logits - torch.max(logits, 1)[0][:, None]
|
|
|
|
logits = logits * self.temperature
|
2023-04-09 16:23:16 +00:00
|
|
|
return logits
|
|
|
|
|
|
|
|
|
|
|
|
class CLUBSample(nn.Module): # Sampled version of the CLUB estimator
|
|
|
|
def __init__(self, last_states, current_states, negative_current_states, predicted_current_states):
|
|
|
|
super(CLUBSample, self).__init__()
|
|
|
|
self.last_states = last_states
|
|
|
|
self.current_states = current_states
|
|
|
|
self.negative_current_states = negative_current_states
|
|
|
|
self.predicted_current_states = predicted_current_states
|
|
|
|
|
|
|
|
def get_mu_var_samples(self, state_dict):
|
|
|
|
dist = state_dict["distribution"]
|
2023-04-12 07:34:11 +00:00
|
|
|
sample = state_dict["sample"] #dist.sample() # Use state_dict["sample"] if you want to use the same sample for all the losses
|
2023-04-09 16:23:16 +00:00
|
|
|
mu = dist.mean
|
|
|
|
var = dist.variance
|
|
|
|
return mu, var, sample
|
|
|
|
|
|
|
|
def loglikeli(self):
|
|
|
|
_, _, pred_sample = self.get_mu_var_samples(self.predicted_current_states)
|
|
|
|
mu_curr, var_curr, _ = self.get_mu_var_samples(self.current_states)
|
2023-04-12 07:34:11 +00:00
|
|
|
logvar_curr = torch.log(var_curr)
|
2023-04-09 16:23:16 +00:00
|
|
|
return (-(mu_curr - pred_sample)**2 /var_curr-logvar_curr).sum(dim=1).mean(dim=0)
|
|
|
|
|
|
|
|
def forward(self):
|
|
|
|
_, _, pred_sample = self.get_mu_var_samples(self.predicted_current_states)
|
|
|
|
mu_curr, var_curr, _ = self.get_mu_var_samples(self.current_states)
|
|
|
|
mu_neg, var_neg, _ = self.get_mu_var_samples(self.negative_current_states)
|
|
|
|
|
2023-04-12 07:34:11 +00:00
|
|
|
sample_size = pred_sample.shape[0]
|
|
|
|
random_index = torch.randperm(sample_size).long()
|
|
|
|
|
2023-04-09 16:23:16 +00:00
|
|
|
pos = (-(mu_curr - pred_sample)**2 /var_curr).sum(dim=1).mean(dim=0)
|
2023-04-12 07:34:11 +00:00
|
|
|
neg = (-(mu_curr - pred_sample[random_index])**2 /var_curr).sum(dim=1).mean(dim=0)
|
|
|
|
#neg = (-(mu_neg - pred_sample)**2 /var_neg).sum(dim=1).mean(dim=0)
|
2023-04-09 16:23:16 +00:00
|
|
|
upper_bound = pos - neg
|
|
|
|
|
|
|
|
return upper_bound/2
|
|
|
|
|
|
|
|
def learning_loss(self):
|
|
|
|
return - self.loglikeli()
|
|
|
|
|
|
|
|
|
|
|
|
if "__name__ == __main__":
|
|
|
|
pass
|