Initial push of DQN agent trained on SpaceInvaders for certification

README.md

@@ -16,7 +16,7 @@ model-index:
       type: SpaceInvadersNoFrameskip-v4
     metrics:
     - type: mean_reward
-      value: 229.50 +/- 143.45
+      value: 546.00 +/- 236.58
       name: mean_reward
       verified: false
 ---
@@ -64,16 +64,16 @@ python -m rl_zoo3.push_to_hub --algo dqn --env SpaceInvadersNoFrameskip-v4 -f lo
 ## Hyperparameters
 ```python
 OrderedDict([('batch_size', 32),
-             ('buffer_size', 50000),
+             ('buffer_size', 75000),
              ('env_wrapper',
               ['stable_baselines3.common.atari_wrappers.AtariWrapper']),
              ('exploration_final_eps', 0.01),
              ('exploration_fraction', 0.1),
              ('frame_stack', 4),
              ('gradient_steps', 1),
-             ('learning_rate', 0.0001),
-             ('learning_starts', 10000),
-             ('n_timesteps', 200000.0),
+             ('learning_rate', 0.00025),
+             ('learning_starts', 20000),
+             ('n_timesteps', 500000.0),
              ('optimize_memory_usage', False),
              ('policy', 'CnnPolicy'),
              ('target_update_interval', 1000),

args.yml

@@ -56,7 +56,7 @@
   - - save_replay_buffer
     - false
   - - seed
-    - 3127200579
+    - 754053280
   - - storage
     - null
   - - study_name

config.yml

@@ -2,7 +2,7 @@
 - - - batch_size
     - 32
   - - buffer_size
-    - 50000
+    - 75000
   - - env_wrapper
     - - stable_baselines3.common.atari_wrappers.AtariWrapper
   - - exploration_final_eps
@@ -14,11 +14,11 @@
   - - gradient_steps
     - 1
   - - learning_rate
-    - 0.0001
+    - 0.00025
   - - learning_starts
-    - 10000
+    - 20000
   - - n_timesteps
-    - 200000.0
+    - 500000.0
   - - optimize_memory_usage
     - false
   - - policy

dqn-SpaceInvadersNoFrameskip-v4.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a94136e89ac821bacb7de70085a5bb01419397b00d39ab84788266dc7f48686f
-size 27218420
+oid sha256:06f5752f277458ae1c6f81ec6b32f4f79229c8baffe3f96056785dfc9b79590e
+size 27218424

dqn-SpaceInvadersNoFrameskip-v4/policy.optimizer.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:51408601a0e87b621fa1229b564d5577c068b667ff9321db34788afbb8653f2c
+oid sha256:ad88d156e7ac6ef5700a71c90fd9362a830a4f865c67e6715ded1bcf8ea57971
 size 13506172

dqn-SpaceInvadersNoFrameskip-v4/policy.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ec0ab4aac1bcae77cb0b4afee649b010125a561482c750fae5a39eaca1bdac7c
+oid sha256:219ccbfebd7380f1ecfaafb01cf9fa614a88960a1e069242cb4d83e7b783a44f
 size 13505370

results.json

@@ -1 +1 @@
-{"mean_reward": 229.5, "std_reward": 143.44772567036398, "is_deterministic": false, "n_eval_episodes": 10, "eval_datetime": "2025-07-11T05:05:49.037484"}
+{"mean_reward": 546.0, "std_reward": 236.57768280207665, "is_deterministic": false, "n_eval_episodes": 10, "eval_datetime": "2025-07-11T06:18:16.342580"}

train_eval_metrics.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1edc63fcb80b0a735a7ad864348314df02678a0c39dfde538b7f0c41c2a48ed0
-size 9149
+oid sha256:36e0b400c718818ef4c344aae17203d3655172cfd7c70f5199a2aec60718acf9
+size 19218