Finalized factoring out projections

This commit is contained in:
Dominik Moritz Roth 2022-09-03 11:59:16 +02:00
parent 0aeea4e2e5
commit 4532135812
3 changed files with 2 additions and 34 deletions

View File

@ -1,31 +0,0 @@
import torch as th
from torch.distributions.multivariate_normal import _batch_mahalanobis
def mahalanobis_alt(u, v, std):
"""
Stolen from Fabian's Code (Public Version)
"""
delta = u - v
return th.triangular_solve(delta, std, upper=False)[0].pow(2).sum([-2, -1])
def mahalanobis(u, v, chol):
delta = u - v
return _batch_mahalanobis(chol, delta)
def frob_sq(diff, is_spd=False):
# If diff is spd, we can use a (probably) more performant algorithm
if is_spd:
return _frob_sq_spd(diff)
return th.norm(diff, p='fro', dim=tuple(range(1, diff.dim()))).pow(2)
def _frob_sq_spd(diff):
return _batch_trace(diff @ diff)
def _batch_trace(x):
return th.diagonal(x, dim1=-2, dim2=-1).sum(-1)

View File

@ -336,8 +336,7 @@ class PPO(GaussianRolloutCollectorAuxclass, OnPolicyAlgorithm):
policy_loss = surrogate_loss + self.ent_coef * entropy_loss + \
trust_region_loss + self.action_coef * action_loss
import pdb
pdb.set_trace()
pg_losses.append(policy_loss.item())
loss = policy_loss + self.vf_coef * value_loss

View File

@ -28,7 +28,7 @@ def main(env_name='ColumbusCandyland_Aux10-v0', timesteps=1_000_000, showRes=Tru
MlpPolicyPPO,
env,
# KLProjectionLayer(trust_region_coeff=0.01),
projection=KLProjectionLayer(trust_region_coeff=0.01),
projection=WassersteinProjectionLayer(trust_region_coeff=0.01),
policy_kwargs={'dist_kwargs': {'neural_strength': Strength.NONE, 'cov_strength': Strength.DIAG, 'parameterization_type':
ParametrizationType.NONE, 'enforce_positive_type': EnforcePositiveType.ABS, 'prob_squashing_type': ProbSquashingType.NONE}},
verbose=0,