Initial commit

README.md

@@ -16,7 +16,7 @@ model-index:
       type: SpaceInvadersNoFrameskip-v4
     metrics:
     - type: mean_reward
-      value: 191.50 +/- 74.87
+      value: 113.00 +/- 46.59
       name: mean_reward
       verified: false
 ---
@@ -62,21 +62,20 @@ python -m rl_zoo3.push_to_hub --algo dqn --env SpaceInvadersNoFrameskip-v4 -f lo
 
 ## Hyperparameters
 ```python
-OrderedDict([('batch_size', 32),
-             ('buffer_size', 100000),
+OrderedDict([('batch_size', 128),
+             ('buffer_size', 50000),
              ('env_wrapper',
               ['stable_baselines3.common.atari_wrappers.AtariWrapper']),
              ('exploration_final_eps', 0.1),
-             ('exploration_fraction', 1),
-             ('frame_stack', 4),
+             ('exploration_fraction', 0.12),
              ('gamma', 0.99),
-             ('gradient_steps', 1),
-             ('learning_rate', 0.00025),
-             ('learning_starts', 10000),
-             ('n_timesteps', 10000000.0),
-             ('optimize_memory_usage', False),
-             ('policy', 'CnnPolicy'),
-             ('target_update_interval', 10000),
+             ('gradient_steps', -1),
+             ('learning_rate', 0.00063),
+             ('learning_starts', 0),
+             ('n_timesteps', 100000.0),
+             ('policy', 'MlpPolicy'),
+             ('policy_kwargs', 'dict(net_arch=[256, 256])'),
+             ('target_update_interval', 250),
              ('train_freq', 4),
              ('normalize', False)])
 ```

args.yml

@@ -54,7 +54,7 @@
   - - save_replay_buffer
     - false
   - - seed
-    - 125510940
+    - 226175671
   - - storage
     - null
   - - study_name

config.yml

@@ -1,31 +1,29 @@
 !!python/object/apply:collections.OrderedDict
 - - - batch_size
-    - 32
+    - 128
   - - buffer_size
-    - 100000
+    - 50000
   - - env_wrapper
     - - stable_baselines3.common.atari_wrappers.AtariWrapper
   - - exploration_final_eps
     - 0.1
   - - exploration_fraction
-    - 1
-  - - frame_stack
-    - 4
+    - 0.12
   - - gamma
     - 0.99
   - - gradient_steps
-    - 1
+    - -1
   - - learning_rate
-    - 0.00025
+    - 0.00063
   - - learning_starts
-    - 10000
+    - 0
   - - n_timesteps
-    - 10000000.0
-  - - optimize_memory_usage
-    - false
+    - 100000.0
   - - policy
-    - CnnPolicy
+    - MlpPolicy
+  - - policy_kwargs
+    - dict(net_arch=[256, 256])
   - - target_update_interval
-    - 10000
+    - 250
   - - train_freq
     - 4

dqn-SpaceInvadersNoFrameskip-v4.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fbfbd0560c768ff659e1f45d4fea44967c2b860cad079c6bdd61345b580c3f39
-size 27219543
+oid sha256:ee1c4148b97a8844ec9b42e540a7e2c982f860e0da6bb3af90559e8094bf0a0d
+size 30057641

dqn-SpaceInvadersNoFrameskip-v4/_stable_baselines3_version

@@ -1 +1 @@
-2.2.0a8
+2.2.0a9

dqn-SpaceInvadersNoFrameskip-v4/policy.optimizer.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:65d44e628b34c4dbf615cd3837e8c82b900a4de1286dedc4a9c3993cf46bae12
-size 13505852
+oid sha256:d5ef147093360095b6e2df012a5142d73afb394b89279f0cd1c3eb604d03b4dd
+size 14996640

dqn-SpaceInvadersNoFrameskip-v4/policy.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:47c58793ca377bcb93b07a6cab7fe98012ffbe8cff718f718277b594e052be6e
-size 13505178
+oid sha256:ebac17fa66dabb65fd91207960dac03d9234fac13ff1cd3fc139fdc3a0aafdf2
+size 14995762

dqn-SpaceInvadersNoFrameskip-v4/system_info.txt

@@ -1,6 +1,6 @@
 - OS: Linux-5.15.120+-x86_64-with-glibc2.35 # 1 SMP Wed Aug 30 11:19:59 UTC 2023
 - Python: 3.10.12
-- Stable-Baselines3: 2.2.0a8
+- Stable-Baselines3: 2.2.0a9
 - PyTorch: 2.1.0+cu118
 - GPU Enabled: False
 - Numpy: 1.23.5

replay.mp4

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:41acc1f1ad783e994942eca11fbc921a10de14675e412dbc1167580a2740bc0a
-size 273434
+oid sha256:25d8d1c6f1271a484d5bf42a9b51aa5ee826f8117566c9da2d8ec04205fa2cb4
+size 270829

results.json

@@ -1 +1 @@
-{"mean_reward": 191.5, "std_reward": 74.8682175559162, "is_deterministic": false, "n_eval_episodes": 10, "eval_datetime": "2023-10-27T00:22:52.139923"}
+{"mean_reward": 113.0, "std_reward": 46.593991028886975, "is_deterministic": false, "n_eval_episodes": 10, "eval_datetime": "2023-10-27T20:28:41.102187"}

train_eval_metrics.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d543838815a0e5b184008577e9882815cbe08b6bbe32b52c38918beda1295d69
-size 40932
+oid sha256:b56cabe9d758f0493d68e97d2854d2ed9a78df318b0c8545b8917dc1ff4215c9
+size 5651