Instructions to use DHDRL/adaptive-wafer-rl with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- stable-baselines3
How to use DHDRL/adaptive-wafer-rl with stable-baselines3:
from huggingface_sb3 import load_from_hub checkpoint = load_from_hub( repo_id="DHDRL/adaptive-wafer-rl", filename="{MODEL FILENAME}.zip", ) - Notebooks
- Google Colab
- Kaggle
Update gru_env_wrappers.py
Browse files- gru_env_wrappers.py +1 -6
gru_env_wrappers.py
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
#!/usr/bin/env python3
|
| 2 |
"""
|
| 3 |
-
GRU-Aware Environment Wrapper
|
| 4 |
=====================================================
|
| 5 |
"""
|
| 6 |
|
|
@@ -13,9 +13,6 @@ class GRUStateManager(gym.Wrapper):
|
|
| 13 |
"""
|
| 14 |
CRITICAL WRAPPER: Manages GRU hidden state lifecycle externally.
|
| 15 |
|
| 16 |
-
This prevents hidden state from being stored in the model,
|
| 17 |
-
which would cause massive checkpoint files.
|
| 18 |
-
|
| 19 |
Responsibilities:
|
| 20 |
- Store hidden state outside model
|
| 21 |
- Reset state at episode boundaries
|
|
@@ -34,7 +31,6 @@ class GRUStateManager(gym.Wrapper):
|
|
| 34 |
self._episode_count = 0
|
| 35 |
|
| 36 |
def set_policy(self, policy):
|
| 37 |
-
"""Set policy reference after wrapper creation."""
|
| 38 |
self.policy = policy
|
| 39 |
self._initialize_hidden_state()
|
| 40 |
|
|
@@ -96,7 +92,6 @@ class GRUStateManager(gym.Wrapper):
|
|
| 96 |
class CurriculumGRUToggleWrapper(gym.Wrapper):
|
| 97 |
"""
|
| 98 |
Tracks curriculum phase for logging.
|
| 99 |
-
|
| 100 |
GRU toggling is disabled: GRU must be configured statically at
|
| 101 |
model construction time. This wrapper only tracks phase transitions.
|
| 102 |
"""
|
|
|
|
| 1 |
#!/usr/bin/env python3
|
| 2 |
"""
|
| 3 |
+
GRU-Aware Environment Wrapper
|
| 4 |
=====================================================
|
| 5 |
"""
|
| 6 |
|
|
|
|
| 13 |
"""
|
| 14 |
CRITICAL WRAPPER: Manages GRU hidden state lifecycle externally.
|
| 15 |
|
|
|
|
|
|
|
|
|
|
| 16 |
Responsibilities:
|
| 17 |
- Store hidden state outside model
|
| 18 |
- Reset state at episode boundaries
|
|
|
|
| 31 |
self._episode_count = 0
|
| 32 |
|
| 33 |
def set_policy(self, policy):
|
|
|
|
| 34 |
self.policy = policy
|
| 35 |
self._initialize_hidden_state()
|
| 36 |
|
|
|
|
| 92 |
class CurriculumGRUToggleWrapper(gym.Wrapper):
|
| 93 |
"""
|
| 94 |
Tracks curriculum phase for logging.
|
|
|
|
| 95 |
GRU toggling is disabled: GRU must be configured statically at
|
| 96 |
model construction time. This wrapper only tracks phase transitions.
|
| 97 |
"""
|