Comment
charts/episodic_return
charts/episodic_return
losses/entropy, loss/policy_entropy
losses/entropy, loss/policy_entropy
losses/explained_variance, misc/explained_variance
losses/explained_variance, misc/explained_variance
loss/value_loss, losses/value_loss
loss/value_loss, losses/value_loss
losses/clipfrac, loss/clipfrac
losses/clipfrac, loss/clipfrac
losses/approx_kl, loss/approxkl
losses/approx_kl, loss/approxkl
Run set
6
Name
6 visualized
gym_id: bigfish
gym_id: bigfish
2
6
gym_id: bossfight
gym_id: bossfight
2
6
gym_id: starpilot
gym_id: starpilot
2
6
gym_id: Breakout-v5
gym_id: Breakout-v5
1
3
gym_id: BeamRider-v5
gym_id: BeamRider-v5
1
3
gym_id: Pong-v5
gym_id: Pong-v5
1
6
gym_id: MountainCar-v0
gym_id: MountainCar-v0
4
12
gym_id: Acrobot-v1
gym_id: Acrobot-v1
4
12
gym_id: CartPole-v1
gym_id: CartPole-v1
4
12
gym_id: MicrortsRandomEnemyShapedReward3-v1
gym_id: MicrortsRandomEnemyShapedReward3-v1
3
9
State
Notes
User
Tags
Created
Runtime
Sweep
alg
anneal_lr
batch_size
capture_video
clip_coef
clip_vloss
cuda
ent_coef
env
exp_name
gae
gae_lambda
gamma
gym_id
learning_rate
max_grad_norm
minibatch_size
network
norm_adv
num_env
num_envs
num_minibatches
num_steps
num_timesteps
play
reward_scale
save_video_interval
save_video_length
seed
torch_deterministic
total_timesteps
track
update_epochs
vf_coef
wandb_entity
wandb_project_name
aux_batch_size
aux_minibatch_size
beta_clone
e_auxiliary
e_policy
n_aux_grad_accum
n_aux_minibatch
n_iteration
Finished
costa-huang
2d 9h 12m 6s
-
ppo2
false
16384
false
0.2
true
true
0.01
bigfish
["baselines-ppo2-None","ppo_procgen"]
true
0.95
0.999
bigfish
0.0005
0.5
2048
-
true
64
64
8
256
25000000
false
1
0
200
2
true
25000000
true
3
0.5
vwxyzjn
ppo-details
-
-
-
-
-
-
-
-
Finished
costa-huang
2d 8h 33m 4s
-
ppo2
false
16384
false
0.2
true
true
0.01
bossfight
["baselines-ppo2-None","ppo_procgen"]
true
0.95
0.999
bossfight
0.0005
0.5
2048
-
true
64
64
8
256
25000000
false
1
0
200
2
true
25000000
true
3
0.5
vwxyzjn
ppo-details
-
-
-
-
-
-
-
-
Finished
costa-huang
2d 6h 49m 32s
-
ppo2
false
16384
false
0.2
true
true
0.01
starpilot
["baselines-ppo2-None","ppo_procgen"]
true
0.95
0.999
starpilot
0.0005
0.5
2048
-
true
64
64
8
256
25000000
false
1
0
200
2
true
25000000
true
3
0.5
vwxyzjn
ppo-details
-
-
-
-
-
-
-
-
Finished
costa-huang
4h 4m 13s
-
-
true
1024
false
0.1
true
true
0.01
-
ppo_atari_envpool
true
0.95
0.99
Breakout-v5
0.00025
0.5
256
-
true
-
8
4
128
-
-
-
-
-
2
true
10000000
true
4
0.5
vwxyzjn
ppo-details
-
-
-
-
-
-
-
-
Finished
costa-huang
4h 1m 33s
-
-
true
1024
false
0.1
true
true
0.01
-
ppo_atari_envpool
true
0.95
0.99
BeamRider-v5
0.00025
0.5
256
-
true
-
8
4
128
-
-
-
-
-
2
true
10000000
true
4
0.5
vwxyzjn
ppo-details
-
-
-
-
-
-
-
-
Finished
costa-huang
3d 6h 48m 17s
-
-
true
1536
false
0.15
true
true
0.01
-
ppo_atari_envpool
true
0.95
0.99
Pong-v5
0.00025
0.5
256
-
true
-
12
6
128
-
-
-
-
-
2
true
10000000
true
3.5
0.5
vwxyzjn
ppo-details
-
-
-
-
-
-
-
-
Finished
costa-huang
24m 8s
-
ppo2
true
512
false
0.2
true
false
0.01
MountainCar-v0
["baselines-ppo2-mlp","baselines-ppo2-mlp-seperate-networks","ppo","ppo_shared"]
true
0.95
0.99
MountainCar-v0
0.00025
0.5
128
mlp
true
-
4
4
128
500000
false
1
0
200
2
true
500000
true
4
0.5
vwxyzjn
ppo-details
-
-
-
-
-
-
-
-
Finished
costa-huang
36m 48s
-
ppo2
true
512
false
0.2
true
false
0.01
Acrobot-v1
["baselines-ppo2-mlp","baselines-ppo2-mlp-seperate-networks","ppo","ppo_shared"]
true
0.95
0.99
Acrobot-v1
0.00025
0.5
128
mlp
true
-
4
4
128
500000
false
1
0
200
2
true
500000
true
4
0.5
vwxyzjn
ppo-details
-
-
-
-
-
-
-
-
Finished
costa-huang
11d 18h 42m 33s
-
ppo2
true
512
false
0.2
true
false
0.01
CartPole-v1
["baselines-ppo2-mlp","baselines-ppo2-mlp-seperate-networks","ppo","ppo_shared"]
true
0.95
0.99
CartPole-v1
0.00025
0.5
128
mlp
true
4
4
4
128
500000
false
1
0
200
2
true
500000
true
4
0.5
vwxyzjn
ppo-details
-
-
-
-
-
-
-
-
Finished
costa-huang
19d 18h 20s
-
ppo2
true
1024
false
0.1
true
true
0.01
MicrortsRandomEnemyShapedReward3-v1
["baselines-ppo2-cnn_gym_microrts","ppo_multidiscrete","ppo_multidiscrete_mask"]
true
0.95
0.99
MicrortsRandomEnemyShapedReward3-v1
0.00025
0.5
256
cnn_gym_microrts
true
8
8
4
128
2000000
false
1
0
200
2
true
2000000
true
4
0.5
vwxyzjn
ppo-details
-
-
-
-
-
-
-
-
1-10
of 19
The panel above shows various PPO metrics. Feel free to toggle the "eyes" button in the panel to toggle the shown metrics for different environment.
💡
Add a comment
Created with ❤️ on Weights & Biases.
https://wandb.ai/vwxyzjn/ppo-details/reports/Matching-the-metrics--VmlldzoxMzY5OTMy