Skip to content
New issue

Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.

#

By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? # to your account

Remove the unnecessary regular advantage code in PPO #287

Merged
merged 18 commits into from
Oct 4, 2022
Merged

Remove the unnecessary regular advantage code in PPO #287

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
18 commits
Select commit Hold shift + click to select a range
c242e9c
Update ppo.py
bragajj Oct 3, 2022
1770542
Update ppo_atari.py
bragajj Oct 3, 2022
595d48c
Update ppo_atari_envpool.py
bragajj Oct 3, 2022
46a1839
Update ppo_continuous_action.py
bragajj Oct 3, 2022
465dee6
Update ppo_atari_lstm.py
bragajj Oct 3, 2022
fd3665a
Update ppo_pettingzoo_ma_atari.py
bragajj Oct 3, 2022
b0994b7
Update ppo_procgen.py
bragajj Oct 3, 2022
a4dfadf
GAE revisions #207
bragajj Oct 3, 2022
f803312
GAE revisions #207
bragajj Oct 3, 2022
0a7df45
GAE revisions #207
bragajj Oct 3, 2022
54a1d97
GAE revisions #207
bragajj Oct 3, 2022
a35b560
GAE revisions #207
bragajj Oct 3, 2022
b5cf6b6
GAE revisions #207
bragajj Oct 3, 2022
84e20ba
GAE revisions #207
bragajj Oct 3, 2022
e21863c
GAE revisions #207
bragajj Oct 3, 2022
33f0683
GAE revisions #207
bragajj Oct 3, 2022
6cc5efd
GAE revisions #207
bragajj Oct 3, 2022
06c5507
Update ppo_rnd_envpool.py
bragajj Oct 4, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
38 changes: 12 additions & 26 deletions cleanrl/ppo.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,8 +48,6 @@ def parse_args():
help="the number of steps to run in each environment per policy rollout")
parser.add_argument("--anneal-lr", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
help="Toggle learning rate annealing for policy and value networks")
parser.add_argument("--gae", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
help="Use GAE for advantage computation")
parser.add_argument("--gamma", type=float, default=0.99,
help="the discount factor gamma")
parser.add_argument("--gae-lambda", type=float, default=0.95,
Expand Down Expand Up @@ -216,30 +214,18 @@ def get_action_and_value(self, x, action=None):
# bootstrap value if not done
with torch.no_grad():
next_value = agent.get_value(next_obs).reshape(1, -1)
if args.gae:
advantages = torch.zeros_like(rewards).to(device)
lastgaelam = 0
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
nextvalues = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
nextvalues = values[t + 1]
delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
returns = advantages + values
else:
returns = torch.zeros_like(rewards).to(device)
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
next_return = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
next_return = returns[t + 1]
returns[t] = rewards[t] + args.gamma * nextnonterminal * next_return
advantages = returns - values
advantages = torch.zeros_like(rewards).to(device)
lastgaelam = 0
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
nextvalues = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
nextvalues = values[t + 1]
delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
returns = advantages + values

# flatten the batch
b_obs = obs.reshape((-1,) + envs.single_observation_space.shape)
Expand Down
38 changes: 12 additions & 26 deletions cleanrl/ppo_atari.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -55,8 +55,6 @@ def parse_args():
help="the number of steps to run in each environment per policy rollout")
parser.add_argument("--anneal-lr", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
help="Toggle learning rate annealing for policy and value networks")
parser.add_argument("--gae", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
help="Use GAE for advantage computation")
parser.add_argument("--gamma", type=float, default=0.99,
help="the discount factor gamma")
parser.add_argument("--gae-lambda", type=float, default=0.95,
Expand Down Expand Up @@ -232,30 +230,18 @@ def get_action_and_value(self, x, action=None):
# bootstrap value if not done
with torch.no_grad():
next_value = agent.get_value(next_obs).reshape(1, -1)
if args.gae:
advantages = torch.zeros_like(rewards).to(device)
lastgaelam = 0
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
nextvalues = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
nextvalues = values[t + 1]
delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
returns = advantages + values
else:
returns = torch.zeros_like(rewards).to(device)
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
next_return = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
next_return = returns[t + 1]
returns[t] = rewards[t] + args.gamma * nextnonterminal * next_return
advantages = returns - values
advantages = torch.zeros_like(rewards).to(device)
lastgaelam = 0
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
nextvalues = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
nextvalues = values[t + 1]
delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
returns = advantages + values

# flatten the batch
b_obs = obs.reshape((-1,) + envs.single_observation_space.shape)
Expand Down
38 changes: 12 additions & 26 deletions cleanrl/ppo_atari_envpool.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -49,8 +49,6 @@ def parse_args():
help="the number of steps to run in each environment per policy rollout")
parser.add_argument("--anneal-lr", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
help="Toggle learning rate annealing for policy and value networks")
parser.add_argument("--gae", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
help="Use GAE for advantage computation")
parser.add_argument("--gamma", type=float, default=0.99,
help="the discount factor gamma")
parser.add_argument("--gae-lambda", type=float, default=0.95,
Expand Down Expand Up @@ -259,30 +257,18 @@ def get_action_and_value(self, x, action=None):
# bootstrap value if not done
with torch.no_grad():
next_value = agent.get_value(next_obs).reshape(1, -1)
if args.gae:
advantages = torch.zeros_like(rewards).to(device)
lastgaelam = 0
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
nextvalues = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
nextvalues = values[t + 1]
delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
returns = advantages + values
else:
returns = torch.zeros_like(rewards).to(device)
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
next_return = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
next_return = returns[t + 1]
returns[t] = rewards[t] + args.gamma * nextnonterminal * next_return
advantages = returns - values
advantages = torch.zeros_like(rewards).to(device)
lastgaelam = 0
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
nextvalues = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
nextvalues = values[t + 1]
delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
returns = advantages + values

# flatten the batch
b_obs = obs.reshape((-1,) + envs.single_observation_space.shape)
Expand Down
38 changes: 12 additions & 26 deletions cleanrl/ppo_atari_lstm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -55,8 +55,6 @@ def parse_args():
help="the number of steps to run in each environment per policy rollout")
parser.add_argument("--anneal-lr", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
help="Toggle learning rate annealing for policy and value networks")
parser.add_argument("--gae", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
help="Use GAE for advantage computation")
parser.add_argument("--gamma", type=float, default=0.99,
help="the discount factor gamma")
parser.add_argument("--gae-lambda", type=float, default=0.95,
Expand Down Expand Up @@ -268,30 +266,18 @@ def get_action_and_value(self, x, lstm_state, done, action=None):
next_lstm_state,
next_done,
).reshape(1, -1)
if args.gae:
advantages = torch.zeros_like(rewards).to(device)
lastgaelam = 0
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
nextvalues = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
nextvalues = values[t + 1]
delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
returns = advantages + values
else:
returns = torch.zeros_like(rewards).to(device)
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
next_return = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
next_return = returns[t + 1]
returns[t] = rewards[t] + args.gamma * nextnonterminal * next_return
advantages = returns - values
advantages = torch.zeros_like(rewards).to(device)
lastgaelam = 0
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
nextvalues = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
nextvalues = values[t + 1]
delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
returns = advantages + values

# flatten the batch
b_obs = obs.reshape((-1,) + envs.single_observation_space.shape)
Expand Down
38 changes: 12 additions & 26 deletions cleanrl/ppo_atari_multigpu.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,8 +57,6 @@ def parse_args():
help="the number of steps to run in each environment per policy rollout")
parser.add_argument("--anneal-lr", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
help="Toggle learning rate annealing for policy and value networks")
parser.add_argument("--gae", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
help="Use GAE for advantage computation")
parser.add_argument("--gamma", type=float, default=0.99,
help="the discount factor gamma")
parser.add_argument("--gae-lambda", type=float, default=0.95,
Expand Down Expand Up @@ -274,30 +272,18 @@ def get_action_and_value(self, x, action=None):
# bootstrap value if not done
with torch.no_grad():
next_value = agent.get_value(next_obs).reshape(1, -1)
if args.gae:
advantages = torch.zeros_like(rewards).to(device)
lastgaelam = 0
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
nextvalues = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
nextvalues = values[t + 1]
delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
returns = advantages + values
else:
returns = torch.zeros_like(rewards).to(device)
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
next_return = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
next_return = returns[t + 1]
returns[t] = rewards[t] + args.gamma * nextnonterminal * next_return
advantages = returns - values
advantages = torch.zeros_like(rewards).to(device)
lastgaelam = 0
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
nextvalues = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
nextvalues = values[t + 1]
delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
returns = advantages + values

# flatten the batch
b_obs = obs.reshape((-1,) + envs.single_observation_space.shape)
Expand Down
38 changes: 12 additions & 26 deletions cleanrl/ppo_continuous_action.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,8 +48,6 @@ def parse_args():
help="the number of steps to run in each environment per policy rollout")
parser.add_argument("--anneal-lr", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
help="Toggle learning rate annealing for policy and value networks")
parser.add_argument("--gae", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
help="Use GAE for advantage computation")
parser.add_argument("--gamma", type=float, default=0.99,
help="the discount factor gamma")
parser.add_argument("--gae-lambda", type=float, default=0.95,
Expand Down Expand Up @@ -224,30 +222,18 @@ def get_action_and_value(self, x, action=None):
# bootstrap value if not done
with torch.no_grad():
next_value = agent.get_value(next_obs).reshape(1, -1)
if args.gae:
advantages = torch.zeros_like(rewards).to(device)
lastgaelam = 0
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
nextvalues = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
nextvalues = values[t + 1]
delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
returns = advantages + values
else:
returns = torch.zeros_like(rewards).to(device)
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
next_return = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
next_return = returns[t + 1]
returns[t] = rewards[t] + args.gamma * nextnonterminal * next_return
advantages = returns - values
advantages = torch.zeros_like(rewards).to(device)
lastgaelam = 0
for t in reversed(range(args.num_steps)):
if t == args.num_steps - 1:
nextnonterminal = 1.0 - next_done
nextvalues = next_value
else:
nextnonterminal = 1.0 - dones[t + 1]
nextvalues = values[t + 1]
delta = rewards[t] + args.gamma * nextvalues * nextnonterminal - values[t]
advantages[t] = lastgaelam = delta + args.gamma * args.gae_lambda * nextnonterminal * lastgaelam
returns = advantages + values

# flatten the batch
b_obs = obs.reshape((-1,) + envs.single_observation_space.shape)
Expand Down
Loading