Skip to content
New issue

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

Sign up for GitHub

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

Already on GitHub? Sign in to your account

[BugFix] Allow for composite action distributions in PPO/A2C losses #2391

Merged
merged 19 commits into from
Sep 4, 2024
Merged

[BugFix] Allow for composite action distributions in PPO/A2C losses #2391

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
19 commits
Select commit Hold shift + click to select a range
627b673
account for composite distribution
albertbou92 Aug 10, 2024
9b08555
tests
albertbou92 Aug 11, 2024
a668d0c
fix tests ppo
albertbou92 Aug 12, 2024
e65e1a3
fix tests ppo
albertbou92 Aug 12, 2024
fd60ad7
fix tests ppo
albertbou92 Aug 12, 2024
bc00e2d
fix tests ppo
albertbou92 Aug 12, 2024
e3f61f7
fix tests ppo
albertbou92 Aug 12, 2024
4eb413d
fix tests ppo
albertbou92 Aug 12, 2024
560ddad
fix tests ppo
albertbou92 Aug 12, 2024
2865c92
format
albertbou92 Aug 12, 2024
13a092d
a2c tests
albertbou92 Aug 14, 2024
8bc8e03
a2c tests
albertbou92 Aug 14, 2024
07ec262
a2c tests
albertbou92 Aug 14, 2024
a95eb3d
a2c tests
albertbou92 Aug 14, 2024
8f00828
format
albertbou92 Aug 14, 2024
9ce3c58
docstrings and minor fixes
albertbou92 Aug 14, 2024
ad32b90
docstrings and minor fixes
albertbou92 Aug 14, 2024
37b733f
docstrings fix
albertbou92 Aug 14, 2024
69922fa
added required grad for td
albertbou92 Aug 26, 2024
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
Next Next commit
account for composite distribution
  • Loading branch information
@albertbou92
albertbou92 committed Aug 10, 2024
commit 627b673637bebe9d9fe9f30642979ba63bf93def
19 changes: 16 additions & 3 deletions torchrl/objectives/a2c.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,7 +11,12 @@

import torch
from tensordict import TensorDict, TensorDictBase, TensorDictParams
from tensordict.nn import dispatch, ProbabilisticTensorDictSequential, TensorDictModule
from tensordict.nn import (
CompositeDistribution,
dispatch,
ProbabilisticTensorDictSequential,
TensorDictModule,
)
from tensordict.utils import NestedKey
from torch import distributions as d

Expand Down Expand Up @@ -383,7 +388,11 @@ def get_entropy_bonus(self, dist: d.Distribution) -> torch.Tensor:
entropy = dist.entropy()
except NotImplementedError:
x = dist.rsample((self.samples_mc_entropy,))
entropy = -dist.log_prob(x).mean(0)
if isinstance(dist, CompositeDistribution):
log_prob = dist.log_prob(x).get(self.tensor_keys.sample_log_prob)
else:
log_prob = dist.log_prob(x)
entropy = -log_prob.mean(0)
vmoens marked this conversation as resolved.
Show resolved Hide resolved
return entropy.unsqueeze(-1)

def _log_probs(
Expand All @@ -402,7 +411,11 @@ def _log_probs(
self.actor_network
) if self.functional else contextlib.nullcontext():
dist = self.actor_network.get_dist(tensordict_clone)
log_prob = dist.log_prob(action)
if isinstance(dist, CompositeDistribution):
tensordict = dist.log_prob(tensordict)
log_prob = tensordict.get(self.tensor_keys.sample_log_prob)
else:
log_prob = dist.log_prob(action)
vmoens marked this conversation as resolved.
Show resolved Hide resolved
log_prob = log_prob.unsqueeze(-1)
return log_prob, dist

Expand Down
26 changes: 23 additions & 3 deletions torchrl/objectives/ppo.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,6 +14,7 @@
import torch
from tensordict import TensorDict, TensorDictBase, TensorDictParams
from tensordict.nn import (
CompositeDistribution,
dispatch,
ProbabilisticTensorDictModule,
ProbabilisticTensorDictSequential,
Expand Down Expand Up @@ -449,7 +450,11 @@ def get_entropy_bonus(self, dist: d.Distribution) -> torch.Tensor:
entropy = dist.entropy()
except NotImplementedError:
x = dist.rsample((self.samples_mc_entropy,))
entropy = -dist.log_prob(x).mean(0)
if isinstance(dist, CompositeDistribution):
log_prob = dist.log_prob(x).get(self.tensor_keys.sample_log_prob)
else:
log_prob = dist.log_prob(x)
entropy = -log_prob.mean(0)
vmoens marked this conversation as resolved.
Show resolved Hide resolved
return entropy.unsqueeze(-1)

def _log_weight(
Expand All @@ -466,12 +471,17 @@ def _log_weight(
self.actor_network
) if self.functional else contextlib.nullcontext():
dist = self.actor_network.get_dist(tensordict)
log_prob = dist.log_prob(action)

prev_log_prob = tensordict.get(self.tensor_keys.sample_log_prob)
if prev_log_prob.requires_grad:
raise RuntimeError("tensordict prev_log_prob requires grad.")

if isinstance(dist, CompositeDistribution):
vmoens marked this conversation as resolved.
Show resolved Hide resolved
tensordict = dist.log_prob(tensordict)
log_prob = tensordict.get(self.tensor_keys.sample_log_prob)
else:
log_prob = dist.log_prob(action)

log_weight = (log_prob - prev_log_prob).unsqueeze(-1)
kl_approx = (prev_log_prob - log_prob).unsqueeze(-1)

Expand Down Expand Up @@ -1107,7 +1117,17 @@ def forward(self, tensordict: TensorDictBase) -> TensorDict:
kl = torch.distributions.kl.kl_divergence(previous_dist, current_dist)
except NotImplementedError:
x = previous_dist.sample((self.samples_mc_kl,))
kl = (previous_dist.log_prob(x) - current_dist.log_prob(x)).mean(0)
if isinstance(current_dist, CompositeDistribution):
vmoens marked this conversation as resolved.
Show resolved Hide resolved
previous_log_prob = previous_dist.log_prob(x).get(
self.tensor_keys.sample_log_prob
)
current_log_prob = current_dist.log_prob(x).get(
self.tensor_keys.sample_log_prob
)
else:
previous_log_prob = previous_dist.log_prob(x)
current_log_prob = current_dist.log_prob(x)
kl = (previous_log_prob - current_log_prob).mean(0)
kl = kl.unsqueeze(-1)
neg_loss = neg_loss - self.beta * kl
if kl.mean() > self.dtarg * 1.5:
Expand Down