Upload config.json with huggingface_hub

config.json

@@ -4,7 +4,7 @@
   "env": "doom_health_gathering_supreme",
   "experiment": "default_experiment",
   "train_dir": "/home/knight/github/huggingface-deep-rl-course/unit8/train_dir",
-  "restart_behavior": "restart",
+  "restart_behavior": "resume",
   "device": "gpu",
   "seed": null,
   "num_policies": 1,
@@ -16,7 +16,7 @@
   "policy_workers_per_policy": 1,
   "max_policy_lag": 1000,
   "num_workers": 8,
-  "num_envs_per_worker": 8,
+  "num_envs_per_worker": 4,
   "batch_size": 1024,
   "num_batches_per_epoch": 1,
   "num_epochs": 1,
@@ -39,11 +39,11 @@
   "vtrace_rho": 1.0,
   "vtrace_c": 1.0,
   "optimizer": "adam",
-  "adam_eps": 1e-05,
-  "adam_beta1": 0.5,
-  "adam_beta2": 0.9,
-  "max_grad_norm": 0.5,
-  "learning_rate": 3e-05,
+  "adam_eps": 1e-06,
+  "adam_beta1": 0.9,
+  "adam_beta2": 0.999,
+  "max_grad_norm": 4.0,
+  "learning_rate": 0.0001,
   "lr_schedule": "constant",
   "lr_schedule_kl_threshold": 0.008,
   "obs_subtract_mean": 0.0,
@@ -63,7 +63,7 @@
   "summaries_use_frameskip": true,
   "heartbeat_interval": 20,
   "heartbeat_reporting_interval": 600,
-  "train_for_env_steps": 10000000,
+  "train_for_env_steps": 4000000,
   "train_for_seconds": 10000000000,
   "save_every_sec": 120,
   "keep_checkpoints": 2,
@@ -81,7 +81,7 @@
   "encoder_conv_mlp_layers": [
     512
   ],
-  "use_rnn": false,
+  "use_rnn": true,
   "rnn_size": 512,
   "rnn_type": "gru",
   "rnn_num_layers": 1,
@@ -89,7 +89,7 @@
   "nonlinearity": "elu",
   "policy_initialization": "orthogonal",
   "policy_init_gain": 1.0,
-  "actor_critic_share_weights": false,
+  "actor_critic_share_weights": true,
   "adaptive_stddev": true,
   "continuous_tanh_scale": 0.0,
   "initial_stddev": 1.0,
@@ -128,20 +128,12 @@
   "wide_aspect_ratio": false,
   "eval_env_frameskip": 1,
   "fps": 35,
-  "command_line": "--env=doom_health_gathering_supreme --num_workers=8 --num_envs_per_worker=8 --train_for_env_steps=10000000 --restart_behavior=restart --max_grad_norm=0.5 --adam_eps=1e-5 --adam_beta1=0.5 --adam_beta2=0.9 --learning_rate=0.00003 --use_rnn=False --actor_critic_share_weights=False",
+  "command_line": "--env=doom_health_gathering_supreme --num_workers=8 --num_envs_per_worker=4 --train_for_env_steps=4000000",
   "cli_args": {
     "env": "doom_health_gathering_supreme",
-    "restart_behavior": "restart",
     "num_workers": 8,
-    "num_envs_per_worker": 8,
-    "adam_eps": 1e-05,
-    "adam_beta1": 0.5,
-    "adam_beta2": 0.9,
-    "max_grad_norm": 0.5,
-    "learning_rate": 3e-05,
-    "train_for_env_steps": 10000000,
-    "use_rnn": false,
-    "actor_critic_share_weights": false
+    "num_envs_per_worker": 4,
+    "train_for_env_steps": 4000000
   },
   "git_hash": "unknown",
   "git_repo_name": "not a git repository",