Initial commit

README.md

@@ -16,7 +16,7 @@ model-index:
       type: SpaceInvadersNoFrameskip-v4
     metrics:
     - type: mean_reward
-      value: 713.50 +/- 271.45
+      value: 794.00 +/- 247.23
       name: mean_reward
       verified: false
 ---
@@ -62,15 +62,15 @@ python -m rl_zoo3.push_to_hub --algo dqn --env SpaceInvadersNoFrameskip-v4 -f lo
 
 ## Hyperparameters
 ```python
-OrderedDict([('batch_size', 64),
-             ('buffer_size', 150000),
+OrderedDict([('batch_size', 32),
+             ('buffer_size', 100000),
              ('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', 5e-05),
+             ('learning_rate', 0.0001),
              ('learning_starts', 100000),
              ('n_timesteps', 10000000.0),
              ('optimize_memory_usage', False),

args.yml

@@ -1,8 +1,6 @@
 !!python/object/apply:collections.OrderedDict
 - - - algo
     - dqn
-  - - batch_size
-    - -1
   - - conf_file
     - null
   - - device
@@ -18,9 +16,7 @@
   - - gym_packages
     - []
   - - hyperparams
-    - batch_size: 64
-      buffer_size: 150000
-      learning_rate: 5.0e-05
+    - null
   - - log_folder
     - logs/
   - - log_interval
@@ -36,7 +32,7 @@
   - - n_startup_trials
     - 10
   - - n_timesteps
-    - 300000
+    - 1000000
   - - n_trials
     - 500
   - - no_optim_plots
@@ -54,11 +50,11 @@
   - - sampler
     - tpe
   - - save_freq
-    - -1
+    - 100000
   - - save_replay_buffer
     - false
   - - seed
-    - 1976347354
+    - 3242211519
   - - storage
     - null
   - - study_name
@@ -68,7 +64,7 @@
   - - track
     - false
   - - trained_agent
-    - logs/dqn/SpaceInvadersNoFrameskip-v4_14/best_model.zip
+    - ''
   - - truncate_last_trajectory
     - true
   - - uuid

config.yml

@@ -1,8 +1,8 @@
 !!python/object/apply:collections.OrderedDict
 - - - batch_size
-    - 64
+    - 32
   - - buffer_size
-    - 150000
+    - 100000
   - - env_wrapper
     - - stable_baselines3.common.atari_wrappers.AtariWrapper
   - - exploration_final_eps
@@ -14,7 +14,7 @@
   - - gradient_steps
     - 1
   - - learning_rate
-    - 5.0e-05
+    - 0.0001
   - - learning_starts
     - 100000
   - - n_timesteps

dqn-SpaceInvadersNoFrameskip-v4.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8f42db5926a09a365bab2a6bc2141a70662b2f71d4a43c392ce844f2f6ccf050
-size 27224804
+oid sha256:55b69fa4ef2376298d2ce01ce3cfa5ca048da8ff47e53c9e2022a92b16679a5a
+size 27224807

dqn-SpaceInvadersNoFrameskip-v4/policy.optimizer.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6dbbfac6b61c60d3ebba31ab991452ce57f398aed826b2b1b2a4c53672f86ef3
+oid sha256:7a6edc7dc842a574379bede3a7fb4d41843b86641b9248d95b19cf959cc92aa0
 size 13505739

dqn-SpaceInvadersNoFrameskip-v4/policy.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d48c41a27c1fc644c343e762042e7874c076aee5e1669f6099773707b06c7147
+oid sha256:5884d000da089a1d98d2f9284741a6ac57d52037035966483b1dc0f7ebebbc61
 size 13504937

replay.mp4

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:10e6739dbc4fd2f9821608abf08818f6decdccd7a447b56fccf312dd8df5c683
-size 217644
+oid sha256:3e41ea531f1520c05a66684eec10c00f10517eb05cfa19ef4beda5810799ae6e
+size 239758

results.json

@@ -1 +1 @@
-{"mean_reward": 713.5, "std_reward": 271.4502716889412, "is_deterministic": false, "n_eval_episodes": 10, "eval_datetime": "2023-02-21T03:06:46.014286"}
+{"mean_reward": 794.0, "std_reward": 247.2326839234651, "is_deterministic": false, "n_eval_episodes": 10, "eval_datetime": "2023-02-21T03:08:41.705194"}

train_eval_metrics.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b135317befaa660dfadb2e8e77f85efc03b7e730ff79e204c60a80aeb931bd5d
-size 10880
+oid sha256:14d915c56248fdebdce8c0310d15a3aa0f7c7ba9cfecb6f43cb35f65efbb59a9
+size 37873