231 lines
9.3 KiB
Python
Raw Normal View History

Feature/dataclasses (#996) This PR adds strict typing to the output of `update` and `learn` in all policies. This will likely be the last large refactoring PR before the next release (0.6.0, not 1.0.0), so it requires some attention. Several difficulties were encountered on the path to that goal: 1. The policy hierarchy is actually "broken" in the sense that the keys of dicts that were output by `learn` did not follow the same enhancement (inheritance) pattern as the policies. This is a real problem and should be addressed in the near future. Generally, several aspects of the policy design and hierarchy might deserve a dedicated discussion. 2. Each policy needs to be generic in the stats return type, because one might want to extend it at some point and then also extend the stats. Even within the source code base this pattern is necessary in many places. 3. The interaction between learn and update is a bit quirky, we currently handle it by having update modify special field inside TrainingStats, whereas all other fields are handled by learn. 4. The IQM module is a policy wrapper and required a TrainingStatsWrapper. The latter relies on a bunch of black magic. They were addressed by: 1. Live with the broken hierarchy, which is now made visible by bounds in generics. We use type: ignore where appropriate. 2. Make all policies generic with bounds following the policy inheritance hierarchy (which is incorrect, see above). We experimented a bit with nested TrainingStats classes, but that seemed to add more complexity and be harder to understand. Unfortunately, mypy thinks that the code below is wrong, wherefore we have to add `type: ignore` to the return of each `learn` ```python T = TypeVar("T", bound=int) def f() -> T: return 3 ``` 3. See above 4. Write representative tests for the `TrainingStatsWrapper`. Still, the black magic might cause nasty surprises down the line (I am not proud of it)... Closes #933 --------- Co-authored-by: Maximilian Huettenrauch <m.huettenrauch@appliedai.de> Co-authored-by: Michael Panchenko <m.panchenko@appliedai.de>
2023-12-30 11:09:03 +01:00
from dataclasses import dataclass
from typing import Any, Literal, TypeVar, cast
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
import gymnasium as gym
import numpy as np
import torch
from torch.distributions import Independent, Normal
from tianshou.data import Batch, ReplayBuffer
from tianshou.data.types import ObsBatchProtocol, RolloutBatchProtocol
from tianshou.exploration import BaseNoise
from tianshou.policy import DDPGPolicy
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
from tianshou.policy.base import TLearningRateScheduler
Feature/dataclasses (#996) This PR adds strict typing to the output of `update` and `learn` in all policies. This will likely be the last large refactoring PR before the next release (0.6.0, not 1.0.0), so it requires some attention. Several difficulties were encountered on the path to that goal: 1. The policy hierarchy is actually "broken" in the sense that the keys of dicts that were output by `learn` did not follow the same enhancement (inheritance) pattern as the policies. This is a real problem and should be addressed in the near future. Generally, several aspects of the policy design and hierarchy might deserve a dedicated discussion. 2. Each policy needs to be generic in the stats return type, because one might want to extend it at some point and then also extend the stats. Even within the source code base this pattern is necessary in many places. 3. The interaction between learn and update is a bit quirky, we currently handle it by having update modify special field inside TrainingStats, whereas all other fields are handled by learn. 4. The IQM module is a policy wrapper and required a TrainingStatsWrapper. The latter relies on a bunch of black magic. They were addressed by: 1. Live with the broken hierarchy, which is now made visible by bounds in generics. We use type: ignore where appropriate. 2. Make all policies generic with bounds following the policy inheritance hierarchy (which is incorrect, see above). We experimented a bit with nested TrainingStats classes, but that seemed to add more complexity and be harder to understand. Unfortunately, mypy thinks that the code below is wrong, wherefore we have to add `type: ignore` to the return of each `learn` ```python T = TypeVar("T", bound=int) def f() -> T: return 3 ``` 3. See above 4. Write representative tests for the `TrainingStatsWrapper`. Still, the black magic might cause nasty surprises down the line (I am not proud of it)... Closes #933 --------- Co-authored-by: Maximilian Huettenrauch <m.huettenrauch@appliedai.de> Co-authored-by: Michael Panchenko <m.panchenko@appliedai.de>
2023-12-30 11:09:03 +01:00
from tianshou.policy.modelfree.ddpg import DDPGTrainingStats
Feature/dataclasses (#996) This PR adds strict typing to the output of `update` and `learn` in all policies. This will likely be the last large refactoring PR before the next release (0.6.0, not 1.0.0), so it requires some attention. Several difficulties were encountered on the path to that goal: 1. The policy hierarchy is actually "broken" in the sense that the keys of dicts that were output by `learn` did not follow the same enhancement (inheritance) pattern as the policies. This is a real problem and should be addressed in the near future. Generally, several aspects of the policy design and hierarchy might deserve a dedicated discussion. 2. Each policy needs to be generic in the stats return type, because one might want to extend it at some point and then also extend the stats. Even within the source code base this pattern is necessary in many places. 3. The interaction between learn and update is a bit quirky, we currently handle it by having update modify special field inside TrainingStats, whereas all other fields are handled by learn. 4. The IQM module is a policy wrapper and required a TrainingStatsWrapper. The latter relies on a bunch of black magic. They were addressed by: 1. Live with the broken hierarchy, which is now made visible by bounds in generics. We use type: ignore where appropriate. 2. Make all policies generic with bounds following the policy inheritance hierarchy (which is incorrect, see above). We experimented a bit with nested TrainingStats classes, but that seemed to add more complexity and be harder to understand. Unfortunately, mypy thinks that the code below is wrong, wherefore we have to add `type: ignore` to the return of each `learn` ```python T = TypeVar("T", bound=int) def f() -> T: return 3 ``` 3. See above 4. Write representative tests for the `TrainingStatsWrapper`. Still, the black magic might cause nasty surprises down the line (I am not proud of it)... Closes #933 --------- Co-authored-by: Maximilian Huettenrauch <m.huettenrauch@appliedai.de> Co-authored-by: Michael Panchenko <m.panchenko@appliedai.de>
2023-12-30 11:09:03 +01:00
@dataclass
class REDQTrainingStats(DDPGTrainingStats):
"""A data structure for storing loss statistics of the REDQ learn step."""
alpha: float | None = None
alpha_loss: float | None = None
TREDQTrainingStats = TypeVar("TREDQTrainingStats", bound=REDQTrainingStats)
class REDQPolicy(DDPGPolicy[TREDQTrainingStats]):
"""Implementation of REDQ. arXiv:2101.05982.
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
:param actor: The actor network following the rules in
:class:`~tianshou.policy.BasePolicy`. (s -> model_output)
:param actor_optim: The optimizer for actor network.
:param critic: The critic network. (s, a -> Q(s, a))
:param critic_optim: The optimizer for critic network.
:param action_space: Env's action space.
:param ensemble_size: Number of sub-networks in the critic ensemble.
:param subset_size: Number of networks in the subset.
:param tau: Param for soft update of the target network.
:param gamma: Discount factor, in [0, 1].
:param alpha: entropy regularization coefficient.
If a tuple (target_entropy, log_alpha, alpha_optim) is provided, then
alpha is automatically tuned.
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
:param exploration_noise: The exploration noise, added to the action. Defaults
to ``GaussianNoise(sigma=0.1)``.
:param estimation_step: The number of steps to look ahead.
:param actor_delay: Number of critic updates before an actor update.
:param observation_space: Env's observation space.
:param action_scaling: if True, scale the action from [-1, 1] to the range
of action_space. Only used if the action_space is continuous.
:param action_bound_method: method to bound action to range [-1, 1].
Only used if the action_space is continuous.
:param lr_scheduler: if not None, will be called in `policy.update()`.
.. seealso::
Please refer to :class:`~tianshou.policy.BasePolicy` for more detailed
explanation.
"""
def __init__(
self,
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
*,
actor: torch.nn.Module,
actor_optim: torch.optim.Optimizer,
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
critic: torch.nn.Module,
critic_optim: torch.optim.Optimizer,
action_space: gym.spaces.Box,
ensemble_size: int = 10,
subset_size: int = 2,
tau: float = 0.005,
gamma: float = 0.99,
alpha: float | tuple[float, torch.Tensor, torch.optim.Optimizer] = 0.2,
estimation_step: int = 1,
actor_delay: int = 20,
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
exploration_noise: BaseNoise | Literal["default"] | None = None,
deterministic_eval: bool = True,
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
target_mode: Literal["mean", "min"] = "min",
action_scaling: bool = True,
action_bound_method: Literal["clip"] | None = "clip",
observation_space: gym.Space | None = None,
lr_scheduler: TLearningRateScheduler | None = None,
) -> None:
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
if target_mode not in ("min", "mean"):
raise ValueError(f"Unsupported target_mode: {target_mode}")
if not 0 < subset_size <= ensemble_size:
raise ValueError(
f"Invalid choice of ensemble size or subset size. "
f"Should be 0 < {subset_size=} <= {ensemble_size=}",
)
super().__init__(
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
actor=actor,
actor_optim=actor_optim,
critic=critic,
critic_optim=critic_optim,
action_space=action_space,
tau=tau,
gamma=gamma,
exploration_noise=exploration_noise,
estimation_step=estimation_step,
action_scaling=action_scaling,
action_bound_method=action_bound_method,
observation_space=observation_space,
lr_scheduler=lr_scheduler,
)
self.ensemble_size = ensemble_size
self.subset_size = subset_size
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
self.target_mode = target_mode
self.critic_gradient_step = 0
self.actor_delay = actor_delay
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
self.deterministic_eval = deterministic_eval
self.__eps = np.finfo(np.float32).eps.item()
Feature/dataclasses (#996) This PR adds strict typing to the output of `update` and `learn` in all policies. This will likely be the last large refactoring PR before the next release (0.6.0, not 1.0.0), so it requires some attention. Several difficulties were encountered on the path to that goal: 1. The policy hierarchy is actually "broken" in the sense that the keys of dicts that were output by `learn` did not follow the same enhancement (inheritance) pattern as the policies. This is a real problem and should be addressed in the near future. Generally, several aspects of the policy design and hierarchy might deserve a dedicated discussion. 2. Each policy needs to be generic in the stats return type, because one might want to extend it at some point and then also extend the stats. Even within the source code base this pattern is necessary in many places. 3. The interaction between learn and update is a bit quirky, we currently handle it by having update modify special field inside TrainingStats, whereas all other fields are handled by learn. 4. The IQM module is a policy wrapper and required a TrainingStatsWrapper. The latter relies on a bunch of black magic. They were addressed by: 1. Live with the broken hierarchy, which is now made visible by bounds in generics. We use type: ignore where appropriate. 2. Make all policies generic with bounds following the policy inheritance hierarchy (which is incorrect, see above). We experimented a bit with nested TrainingStats classes, but that seemed to add more complexity and be harder to understand. Unfortunately, mypy thinks that the code below is wrong, wherefore we have to add `type: ignore` to the return of each `learn` ```python T = TypeVar("T", bound=int) def f() -> T: return 3 ``` 3. See above 4. Write representative tests for the `TrainingStatsWrapper`. Still, the black magic might cause nasty surprises down the line (I am not proud of it)... Closes #933 --------- Co-authored-by: Maximilian Huettenrauch <m.huettenrauch@appliedai.de> Co-authored-by: Michael Panchenko <m.panchenko@appliedai.de>
2023-12-30 11:09:03 +01:00
self._last_actor_loss = 0.0 # only for logging purposes
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
# TODO: reduce duplication with SACPolicy
self.alpha: float | torch.Tensor
self._is_auto_alpha = not isinstance(alpha, float)
if self._is_auto_alpha:
# TODO: why doesn't mypy understand that this must be a tuple?
alpha = cast(tuple[float, torch.Tensor, torch.optim.Optimizer], alpha)
if alpha[1].shape != torch.Size([1]):
raise ValueError(
f"Expected log_alpha to have shape torch.Size([1]), "
f"but got {alpha[1].shape} instead.",
)
if not alpha[1].requires_grad:
raise ValueError("Expected log_alpha to require gradient, but it doesn't.")
self.target_entropy, self.log_alpha, self.alpha_optim = alpha
self.alpha = self.log_alpha.detach().exp()
else:
# TODO: make mypy undestand this, or switch to something like pyright...
alpha = cast(float, alpha)
self.alpha = alpha
@property
def is_auto_alpha(self) -> bool:
return self._is_auto_alpha
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
# TODO: why override from the base class?
def sync_weight(self) -> None:
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
for o, n in zip(self.critic_old.parameters(), self.critic.parameters(), strict=True):
o.data.copy_(o.data * (1.0 - self.tau) + n.data * self.tau)
def forward( # type: ignore
self,
batch: ObsBatchProtocol,
state: dict | Batch | np.ndarray | None = None,
**kwargs: Any,
) -> Batch:
loc_scale, h = self.actor(batch.obs, state=state, info=batch.info)
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
loc, scale = loc_scale
dist = Independent(Normal(loc, scale), 1)
act = loc if self.deterministic_eval and not self.training else dist.rsample()
log_prob = dist.log_prob(act).unsqueeze(-1)
# apply correction for Tanh squashing when computing logprob from Gaussian
# You can check out the original SAC paper (arXiv 1801.01290): Eq 21.
# in appendix C to get some understanding of this equation.
squashed_action = torch.tanh(act)
log_prob = log_prob - torch.log((1 - squashed_action.pow(2)) + self.__eps).sum(
-1,
keepdim=True,
)
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
return Batch(logits=loc_scale, act=squashed_action, state=h, dist=dist, log_prob=log_prob)
def _target_q(self, buffer: ReplayBuffer, indices: np.ndarray) -> torch.Tensor:
obs_next_batch = Batch(
obs=buffer[indices].obs_next,
info=[None] * len(indices),
) # obs_next: s_{t+n}
obs_next_result = self(obs_next_batch)
a_ = obs_next_result.act
sample_ensemble_idx = np.random.choice(self.ensemble_size, self.subset_size, replace=False)
qs = self.critic_old(obs_next_batch.obs, a_)[sample_ensemble_idx, ...]
if self.target_mode == "min":
target_q, _ = torch.min(qs, dim=0)
elif self.target_mode == "mean":
target_q = torch.mean(qs, dim=0)
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
target_q -= self.alpha * obs_next_result.log_prob
return target_q
Feature/dataclasses (#996) This PR adds strict typing to the output of `update` and `learn` in all policies. This will likely be the last large refactoring PR before the next release (0.6.0, not 1.0.0), so it requires some attention. Several difficulties were encountered on the path to that goal: 1. The policy hierarchy is actually "broken" in the sense that the keys of dicts that were output by `learn` did not follow the same enhancement (inheritance) pattern as the policies. This is a real problem and should be addressed in the near future. Generally, several aspects of the policy design and hierarchy might deserve a dedicated discussion. 2. Each policy needs to be generic in the stats return type, because one might want to extend it at some point and then also extend the stats. Even within the source code base this pattern is necessary in many places. 3. The interaction between learn and update is a bit quirky, we currently handle it by having update modify special field inside TrainingStats, whereas all other fields are handled by learn. 4. The IQM module is a policy wrapper and required a TrainingStatsWrapper. The latter relies on a bunch of black magic. They were addressed by: 1. Live with the broken hierarchy, which is now made visible by bounds in generics. We use type: ignore where appropriate. 2. Make all policies generic with bounds following the policy inheritance hierarchy (which is incorrect, see above). We experimented a bit with nested TrainingStats classes, but that seemed to add more complexity and be harder to understand. Unfortunately, mypy thinks that the code below is wrong, wherefore we have to add `type: ignore` to the return of each `learn` ```python T = TypeVar("T", bound=int) def f() -> T: return 3 ``` 3. See above 4. Write representative tests for the `TrainingStatsWrapper`. Still, the black magic might cause nasty surprises down the line (I am not proud of it)... Closes #933 --------- Co-authored-by: Maximilian Huettenrauch <m.huettenrauch@appliedai.de> Co-authored-by: Michael Panchenko <m.panchenko@appliedai.de>
2023-12-30 11:09:03 +01:00
def learn(self, batch: RolloutBatchProtocol, *args: Any, **kwargs: Any) -> TREDQTrainingStats: # type: ignore
# critic ensemble
weight = getattr(batch, "weight", 1.0)
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
current_qs = self.critic(batch.obs, batch.act).flatten(1)
target_q = batch.returns.flatten()
td = current_qs - target_q
critic_loss = (td.pow(2) * weight).mean()
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
self.critic_optim.zero_grad()
critic_loss.backward()
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
self.critic_optim.step()
batch.weight = torch.mean(td, dim=0) # prio-buffer
self.critic_gradient_step += 1
Feature/dataclasses (#996) This PR adds strict typing to the output of `update` and `learn` in all policies. This will likely be the last large refactoring PR before the next release (0.6.0, not 1.0.0), so it requires some attention. Several difficulties were encountered on the path to that goal: 1. The policy hierarchy is actually "broken" in the sense that the keys of dicts that were output by `learn` did not follow the same enhancement (inheritance) pattern as the policies. This is a real problem and should be addressed in the near future. Generally, several aspects of the policy design and hierarchy might deserve a dedicated discussion. 2. Each policy needs to be generic in the stats return type, because one might want to extend it at some point and then also extend the stats. Even within the source code base this pattern is necessary in many places. 3. The interaction between learn and update is a bit quirky, we currently handle it by having update modify special field inside TrainingStats, whereas all other fields are handled by learn. 4. The IQM module is a policy wrapper and required a TrainingStatsWrapper. The latter relies on a bunch of black magic. They were addressed by: 1. Live with the broken hierarchy, which is now made visible by bounds in generics. We use type: ignore where appropriate. 2. Make all policies generic with bounds following the policy inheritance hierarchy (which is incorrect, see above). We experimented a bit with nested TrainingStats classes, but that seemed to add more complexity and be harder to understand. Unfortunately, mypy thinks that the code below is wrong, wherefore we have to add `type: ignore` to the return of each `learn` ```python T = TypeVar("T", bound=int) def f() -> T: return 3 ``` 3. See above 4. Write representative tests for the `TrainingStatsWrapper`. Still, the black magic might cause nasty surprises down the line (I am not proud of it)... Closes #933 --------- Co-authored-by: Maximilian Huettenrauch <m.huettenrauch@appliedai.de> Co-authored-by: Michael Panchenko <m.panchenko@appliedai.de>
2023-12-30 11:09:03 +01:00
alpha_loss = None
# actor
if self.critic_gradient_step % self.actor_delay == 0:
obs_result = self(batch)
a = obs_result.act
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
current_qa = self.critic(batch.obs, a).mean(dim=0).flatten()
actor_loss = (self.alpha * obs_result.log_prob.flatten() - current_qa).mean()
self.actor_optim.zero_grad()
actor_loss.backward()
self.actor_optim.step()
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
if self.is_auto_alpha:
log_prob = obs_result.log_prob.detach() + self._target_entropy
alpha_loss = -(self._log_alpha * log_prob).mean()
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
self.alpha_optim.zero_grad()
alpha_loss.backward()
Remove kwargs in policy init (#950) Closes #947 This removes all kwargs from all policy constructors. While doing that, I also improved several names and added a whole lot of TODOs. ## Functional changes: 1. Added possibility to pass None as `critic2` and `critic2_optim`. In fact, the default behavior then should cover the absolute majority of cases 2. Added a function called `clone_optimizer` as a temporary measure to support passing `critic2_optim=None` ## Breaking changes: 1. `action_space` is no longer optional. In fact, it already was non-optional, as there was a ValueError in BasePolicy.init. So now several examples were fixed to reflect that 2. `reward_normalization` removed from DDPG and children. It was never allowed to pass it as `True` there, an error would have been raised in `compute_n_step_reward`. Now I removed it from the interface 3. renamed `critic1` and similar to `critic`, in order to have uniform interfaces. Note that the `critic` in DDPG was optional for the sole reason that child classes used `critic1`. I removed this optionality (DDPG can't do anything with `critic=None`) 4. Several renamings of fields (mostly private to public, so backwards compatible) ## Additional changes: 1. Removed type and default declaration from docstring. This kind of duplication is really not necessary 2. Policy constructors are now only called using named arguments, not a fragile mixture of positional and named as before 5. Minor beautifications in typing and code 6. Generally shortened docstrings and made them uniform across all policies (hopefully) ## Comment: With these changes, several problems in tianshou's inheritance hierarchy become more apparent. I tried highlighting them for future work. --------- Co-authored-by: Dominik Jain <d.jain@appliedai.de>
2023-10-08 17:57:03 +02:00
self.alpha_optim.step()
self.alpha = self.log_alpha.detach().exp()
self.sync_weight()
if self.critic_gradient_step % self.actor_delay == 0:
Feature/dataclasses (#996) This PR adds strict typing to the output of `update` and `learn` in all policies. This will likely be the last large refactoring PR before the next release (0.6.0, not 1.0.0), so it requires some attention. Several difficulties were encountered on the path to that goal: 1. The policy hierarchy is actually "broken" in the sense that the keys of dicts that were output by `learn` did not follow the same enhancement (inheritance) pattern as the policies. This is a real problem and should be addressed in the near future. Generally, several aspects of the policy design and hierarchy might deserve a dedicated discussion. 2. Each policy needs to be generic in the stats return type, because one might want to extend it at some point and then also extend the stats. Even within the source code base this pattern is necessary in many places. 3. The interaction between learn and update is a bit quirky, we currently handle it by having update modify special field inside TrainingStats, whereas all other fields are handled by learn. 4. The IQM module is a policy wrapper and required a TrainingStatsWrapper. The latter relies on a bunch of black magic. They were addressed by: 1. Live with the broken hierarchy, which is now made visible by bounds in generics. We use type: ignore where appropriate. 2. Make all policies generic with bounds following the policy inheritance hierarchy (which is incorrect, see above). We experimented a bit with nested TrainingStats classes, but that seemed to add more complexity and be harder to understand. Unfortunately, mypy thinks that the code below is wrong, wherefore we have to add `type: ignore` to the return of each `learn` ```python T = TypeVar("T", bound=int) def f() -> T: return 3 ``` 3. See above 4. Write representative tests for the `TrainingStatsWrapper`. Still, the black magic might cause nasty surprises down the line (I am not proud of it)... Closes #933 --------- Co-authored-by: Maximilian Huettenrauch <m.huettenrauch@appliedai.de> Co-authored-by: Michael Panchenko <m.panchenko@appliedai.de>
2023-12-30 11:09:03 +01:00
self._last_actor_loss = actor_loss.item()
if self.is_auto_alpha:
self.alpha = cast(torch.Tensor, self.alpha)
return REDQTrainingStats( # type: ignore[return-value]
actor_loss=self._last_actor_loss,
critic_loss=critic_loss.item(),
alpha=self.alpha.item() if isinstance(self.alpha, torch.Tensor) else self.alpha,
alpha_loss=alpha_loss,
)