edbeeching HF staff commited on
Commit
381f150
1 Parent(s): f5e1983

Upload cfg.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. cfg.json +132 -0
cfg.json ADDED
@@ -0,0 +1,132 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "help": false,
3
+ "algo": "APPO",
4
+ "env": "mujoco_doublependulum",
5
+ "experiment": "04_mujoco_all_envs_see_1111_env_mujoco_doublependulum",
6
+ "train_dir": "./train_dir/envpool_mujoco_all_envs/mujoco_all_envs_0",
7
+ "restart_behavior": "resume",
8
+ "device": "gpu",
9
+ "seed": 1111,
10
+ "num_policies": 1,
11
+ "async_rl": true,
12
+ "serial_mode": false,
13
+ "batched_sampling": true,
14
+ "num_batches_to_accumulate": 2,
15
+ "worker_num_splits": 1,
16
+ "policy_workers_per_policy": 1,
17
+ "max_policy_lag": 1000,
18
+ "num_workers": 8,
19
+ "num_envs_per_worker": 1,
20
+ "batch_size": 1024,
21
+ "num_batches_per_epoch": 4,
22
+ "num_epochs": 2,
23
+ "rollout": 64,
24
+ "recurrence": 1,
25
+ "shuffle_minibatches": false,
26
+ "gamma": 0.99,
27
+ "reward_scale": 1,
28
+ "reward_clip": 1000.0,
29
+ "value_bootstrap": false,
30
+ "normalize_returns": true,
31
+ "exploration_loss_coeff": 0.0,
32
+ "value_loss_coeff": 1.3,
33
+ "kl_loss_coeff": 0.1,
34
+ "exploration_loss": "entropy",
35
+ "gae_lambda": 0.95,
36
+ "ppo_clip_ratio": 0.2,
37
+ "ppo_clip_value": 1.0,
38
+ "with_vtrace": false,
39
+ "vtrace_rho": 1.0,
40
+ "vtrace_c": 1.0,
41
+ "optimizer": "adam",
42
+ "adam_eps": 1e-06,
43
+ "adam_beta1": 0.9,
44
+ "adam_beta2": 0.999,
45
+ "max_grad_norm": 3.5,
46
+ "learning_rate": 0.00295,
47
+ "lr_schedule": "linear_decay",
48
+ "lr_schedule_kl_threshold": 0.008,
49
+ "obs_subtract_mean": 0.0,
50
+ "obs_scale": 1.0,
51
+ "normalize_input": true,
52
+ "normalize_input_keys": null,
53
+ "decorrelate_experience_max_seconds": 0,
54
+ "decorrelate_envs_on_one_worker": true,
55
+ "actor_worker_gpus": [],
56
+ "set_workers_cpu_affinity": true,
57
+ "force_envs_single_thread": false,
58
+ "default_niceness": 0,
59
+ "log_to_file": true,
60
+ "experiment_summaries_interval": 3,
61
+ "flush_summaries_interval": 30,
62
+ "stats_avg": 100,
63
+ "summaries_use_frameskip": true,
64
+ "heartbeat_interval": 10,
65
+ "heartbeat_reporting_interval": 60,
66
+ "train_for_env_steps": 50000000,
67
+ "train_for_seconds": 10000000000,
68
+ "save_every_sec": 15,
69
+ "keep_checkpoints": 2,
70
+ "load_checkpoint_kind": "latest",
71
+ "save_milestones_sec": -1,
72
+ "save_best_every_sec": 5,
73
+ "save_best_metric": "reward",
74
+ "save_best_after": 100000,
75
+ "benchmark": false,
76
+ "encoder_mlp_layers": [
77
+ 64,
78
+ 64
79
+ ],
80
+ "encoder_conv_architecture": "convnet_simple",
81
+ "encoder_conv_mlp_layers": [
82
+ 512
83
+ ],
84
+ "use_rnn": false,
85
+ "rnn_size": 512,
86
+ "rnn_type": "gru",
87
+ "rnn_num_layers": 1,
88
+ "decoder_mlp_layers": [],
89
+ "nonlinearity": "tanh",
90
+ "policy_initialization": "torch_default",
91
+ "policy_init_gain": 1.0,
92
+ "actor_critic_share_weights": true,
93
+ "adaptive_stddev": false,
94
+ "continuous_tanh_scale": 0.0,
95
+ "initial_stddev": 1.0,
96
+ "use_env_info_cache": false,
97
+ "env_gpu_actions": false,
98
+ "env_frameskip": 1,
99
+ "env_framestack": 1,
100
+ "pixel_format": "CHW",
101
+ "use_record_episode_statistics": false,
102
+ "with_wandb": false,
103
+ "wandb_user": null,
104
+ "wandb_project": "sample_factory",
105
+ "wandb_group": null,
106
+ "wandb_job_type": "SF",
107
+ "wandb_tags": [],
108
+ "with_pbt": false,
109
+ "pbt_mix_policies_in_one_env": true,
110
+ "pbt_period_env_steps": 5000000,
111
+ "pbt_start_mutation": 20000000,
112
+ "pbt_replace_fraction": 0.3,
113
+ "pbt_mutation_rate": 0.15,
114
+ "pbt_replace_reward_gap": 0.1,
115
+ "pbt_replace_reward_gap_absolute": 1e-06,
116
+ "pbt_optimize_gamma": false,
117
+ "pbt_target_objective": "true_objective",
118
+ "pbt_perturb_min": 1.1,
119
+ "pbt_perturb_max": 1.5,
120
+ "env_agents": 256,
121
+ "command_line": "--algo=APPO --with_wandb=False --seed=1111 --env=mujoco_doublependulum --experiment=04_mujoco_all_envs_see_1111_env_mujoco_doublependulum --train_dir=./train_dir/envpool_mujoco_all_envs/mujoco_all_envs_0",
122
+ "cli_args": {
123
+ "algo": "APPO",
124
+ "env": "mujoco_doublependulum",
125
+ "experiment": "04_mujoco_all_envs_see_1111_env_mujoco_doublependulum",
126
+ "train_dir": "./train_dir/envpool_mujoco_all_envs/mujoco_all_envs_0",
127
+ "seed": 1111,
128
+ "with_wandb": false
129
+ },
130
+ "git_hash": "9da68b57eecd73c3c884c1be2d938b46aa7a7f49",
131
+ "git_repo_name": "git@github.com:alex-petrenko/sample-factory.git"
132
+ }