Upload config.json with huggingface_hub

config.json

@@ -2,7 +2,7 @@
   "help": false,
   "algo": "APPO",
   "env": "doom_health_gathering_supreme",
-  "experiment": "vizdoom_doom_health_gathering_supreme_gru_2222_1_decorrelate_10M",
+  "experiment": "vizdoom_doom_health_gathering_supreme_lstm_2222_3_decorrelate_10M",
   "train_dir": "/home/labicquette/wip/hugRL/train_dir",
   "restart_behavior": "resume",
   "device": "gpu",
@@ -16,7 +16,7 @@
   "policy_workers_per_policy": 1,
   "max_policy_lag": 1000,
   "num_workers": 12,
-  "num_envs_per_worker": 8,
+  "num_envs_per_worker": 12,
   "batch_size": 2048,
   "num_batches_per_epoch": 1,
   "num_epochs": 1,
@@ -42,7 +42,7 @@
   "adam_eps": 1e-06,
   "adam_beta1": 0.9,
   "adam_beta2": 0.999,
-  "max_grad_norm": 4.0,
+  "max_grad_norm": 0.0,
   "learning_rate": 0.0001,
   "lr_schedule": "constant",
   "lr_schedule_kl_threshold": 0.008,
@@ -61,8 +61,8 @@
   "flush_summaries_interval": 30,
   "stats_avg": 100,
   "summaries_use_frameskip": true,
-  "heartbeat_interval": 20,
-  "heartbeat_reporting_interval": 600,
+  "heartbeat_interval": 10,
+  "heartbeat_reporting_interval": 300,
   "train_for_env_steps": 10000000,
   "train_for_seconds": 10000000000,
   "save_every_sec": 120,
@@ -83,10 +83,10 @@
   ],
   "use_rnn": true,
   "rnn_size": 512,
-  "rnn_type": "gru",
+  "rnn_type": "lstm",
   "rnn_num_layers": 1,
   "decoder_mlp_layers": [],
-  "nonlinearity": "elu",
+  "nonlinearity": "relu",
   "policy_initialization": "orthogonal",
   "policy_init_gain": 1.0,
   "actor_critic_share_weights": true,
@@ -95,7 +95,7 @@
   "initial_stddev": 1.0,
   "use_env_info_cache": false,
   "env_gpu_actions": false,
-  "env_gpu_observations": false,
+  "env_gpu_observations": true,
   "env_frameskip": 4,
   "env_framestack": 1,
   "pixel_format": "CHW",
@@ -128,20 +128,22 @@
   "wide_aspect_ratio": false,
   "eval_env_frameskip": 1,
   "fps": 35,
-  "command_line": "--env=doom_health_gathering_supreme --experiment=vizdoom_doom_health_gathering_supreme_gru_2222_1_decorrelate_10M --num_workers=12 --num_envs_per_worker=8 --train_for_env_steps=10000000 --batch_size=2048 --max_grad_norm=4.0 --nonlinearity=elu --rnn_type=gru --decorrelate_experience_max_seconds=1 --env_gpu_observations=false --seed=2222",
+  "command_line": "--env=doom_health_gathering_supreme --experiment=vizdoom_doom_health_gathering_supreme_lstm_2222_3_decorrelate_10M --num_workers=12 --num_envs_per_worker=12 --train_for_env_steps=10000000 --batch_size=2048 --max_grad_norm=0.0 --nonlinearity=relu --rnn_type=lstm --decorrelate_experience_max_seconds=1 --env_gpu_observations=true --seed=2222 --heartbeat_interval=10 --heartbeat_reporting_interval=300",
   "cli_args": {
     "env": "doom_health_gathering_supreme",
-    "experiment": "vizdoom_doom_health_gathering_supreme_gru_2222_1_decorrelate_10M",
+    "experiment": "vizdoom_doom_health_gathering_supreme_lstm_2222_3_decorrelate_10M",
     "seed": 2222,
     "num_workers": 12,
-    "num_envs_per_worker": 8,
+    "num_envs_per_worker": 12,
     "batch_size": 2048,
-    "max_grad_norm": 4.0,
+    "max_grad_norm": 0.0,
     "decorrelate_experience_max_seconds": 1,
+    "heartbeat_interval": 10,
+    "heartbeat_reporting_interval": 300,
     "train_for_env_steps": 10000000,
-    "rnn_type": "gru",
-    "nonlinearity": "elu",
-    "env_gpu_observations": false
+    "rnn_type": "lstm",
+    "nonlinearity": "relu",
+    "env_gpu_observations": true
   },
   "git_hash": "unknown",
   "git_repo_name": "not a git repository",