[16:39:32] ============================================================
[16:39:32] Exp 20: Parallel DummyVecEnv — 450k steps (sim rebuild + progress fix)
[16:39:32]   Sim 1: localhost:9091 → generated_track
[16:39:32]   Sim 2: localhost:9093 → mountain_track
[16:39:32]   throttle_min=0.2, lr=0.000725, total=450,000
[16:39:32]   Reward: v6 + exploit fix (window=200, min_lap=12.0s)
[16:39:32]   Stuck termination: 40 steps (~2s), hard cap 30s
[16:39:32]   Progress patience: 150 steps (~7.5s at 20fps)
[16:39:32]   Checkpoints: every 20,000 steps
[16:39:32] ============================================================
[16:39:32] Creating DummyVecEnv with two tracks...
starting DonkeyGym env
Setting default: start_delay 5.0
Setting default: max_cte 8.0
Setting default: frame_skip 1
Setting default: cam_resolution (120, 160, 3)
Setting default: log_level 20
Setting default: steer_limit 1.0
Setting default: throttle_min 0.0
Setting default: throttle_max 1.0
loading scene generated_track
starting DonkeyGym env
Setting default: start_delay 5.0
Setting default: max_cte 8.0
Setting default: frame_skip 1
Setting default: cam_resolution (120, 160, 3)
Setting default: log_level 20
Setting default: steer_limit 1.0
Setting default: throttle_min 0.0
Setting default: throttle_max 1.0
loading scene mountain_track
[16:39:34]   VecEnv num_envs=2, obs=(3, 120, 160)
Using cpu device
[16:39:38] PPO created. Starting training...
-----------------------------
| time/              |      |
|    fps             | 28   |
|    iterations      | 1    |
|    time_elapsed    | 142  |
|    total_timesteps | 4096 |
-----------------------------
-----------------------------------------
| time/                   |             |
|    fps                  | 19          |
|    iterations           | 2           |
|    time_elapsed         | 428         |
|    total_timesteps      | 8192        |
| train/                  |             |
|    approx_kl            | 0.106491014 |
|    clip_fraction        | 0.241       |
|    clip_range           | 0.2         |
|    entropy_loss         | -2.81       |
|    explained_variance   | 0.0458      |
|    learning_rate        | 0.000725    |
|    loss                 | -0.0816     |
|    n_updates            | 10          |
|    policy_gradient_loss | -0.0458     |
|    std                  | 0.953       |
|    value_loss           | 1.39        |
-----------------------------------------
----------------------------------------
| time/                   |            |
|    fps                  | 17         |
|    iterations           | 3          |
|    time_elapsed         | 703        |
|    total_timesteps      | 12288      |
| train/                  |            |
|    approx_kl            | 0.15251182 |
|    clip_fraction        | 0.428      |
|    clip_range           | 0.2        |
|    entropy_loss         | -2.74      |
|    explained_variance   | 0.0179     |
|    learning_rate        | 0.000725   |
|    loss                 | 0.0731     |
|    n_updates            | 20         |
|    policy_gradient_loss | -0.0196    |
|    std                  | 0.946      |
|    value_loss           | 18.6       |
----------------------------------------
----------------------------------------
| time/                   |            |
|    fps                  | 17         |
|    iterations           | 4          |
|    time_elapsed         | 955        |
|    total_timesteps      | 16384      |
| train/                  |            |
|    approx_kl            | 0.22289596 |
|    clip_fraction        | 0.505      |
|    clip_range           | 0.2        |
|    entropy_loss         | -2.68      |
|    explained_variance   | -0.418     |
|    learning_rate        | 0.000725   |
|    loss                 | -0.0581    |
|    n_updates            | 30         |
|    policy_gradient_loss | -0.0692    |
|    std                  | 0.882      |
|    value_loss           | 0.303      |
----------------------------------------
----------------------------------------
| time/                   |            |
|    fps                  | 17         |
|    iterations           | 5          |
|    time_elapsed         | 1143       |
|    total_timesteps      | 20480      |
| train/                  |            |
|    approx_kl            | 0.50940317 |
|    clip_fraction        | 0.619      |
|    clip_range           | 0.2        |
|    entropy_loss         | -2.51      |
|    explained_variance   | 0.405      |
|    learning_rate        | 0.000725   |
|    loss                 | -0.117     |
|    n_updates            | 40         |
|    policy_gradient_loss | -0.0885    |
|    std                  | 0.807      |
|    value_loss           | 0.196      |
----------------------------------------
[17:00:28] [20,000/450,000] Checkpoint saved: /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/exp20-parallel-450k-v5/checkpoint_0020000.zip
[17:00:34]   Eval: gen_track=6.7r/67s ❌@67  mountain=5.0r/67s ❌@67
[17:00:34]   NEW BEST: 11.7 combined reward
------------------------------
| time/              |       |
|    fps             | 44    |
|    iterations      | 1     |
|    time_elapsed    | 91    |
|    total_timesteps | 24576 |
------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 31        |
|    iterations           | 2         |
|    time_elapsed         | 260       |
|    total_timesteps      | 28672     |
| train/                  |           |
|    approx_kl            | 0.4478566 |
|    clip_fraction        | 0.658     |
|    clip_range           | 0.2       |
|    entropy_loss         | -2.14     |
|    explained_variance   | 0.597     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0885   |
|    n_updates            | 60        |
|    policy_gradient_loss | -0.0843   |
|    std                  | 0.67      |
|    value_loss           | 0.212     |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 29        |
|    iterations           | 3         |
|    time_elapsed         | 420       |
|    total_timesteps      | 32768     |
| train/                  |           |
|    approx_kl            | 0.9769267 |
|    clip_fraction        | 0.658     |
|    clip_range           | 0.2       |
|    entropy_loss         | -1.96     |
|    explained_variance   | 0.399     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.107    |
|    n_updates            | 70        |
|    policy_gradient_loss | -0.0829   |
|    std                  | 0.61      |
|    value_loss           | 0.403     |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 28        |
|    iterations           | 4         |
|    time_elapsed         | 576       |
|    total_timesteps      | 36864     |
| train/                  |           |
|    approx_kl            | 0.5094907 |
|    clip_fraction        | 0.682     |
|    clip_range           | 0.2       |
|    entropy_loss         | -1.81     |
|    explained_variance   | 0.45      |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0843   |
|    n_updates            | 80        |
|    policy_gradient_loss | -0.076    |
|    std                  | 0.562     |
|    value_loss           | 0.396     |
---------------------------------------
----------------------------------------
| time/                   |            |
|    fps                  | 27         |
|    iterations           | 5          |
|    time_elapsed         | 731        |
|    total_timesteps      | 40960      |
| train/                  |            |
|    approx_kl            | 0.70720506 |
|    clip_fraction        | 0.703      |
|    clip_range           | 0.2        |
|    entropy_loss         | -1.61      |
|    explained_variance   | 0.466      |
|    learning_rate        | 0.000725   |
|    loss                 | -0.0856    |
|    n_updates            | 90         |
|    policy_gradient_loss | -0.0797    |
|    std                  | 0.513      |
|    value_loss           | 0.468      |
----------------------------------------
[17:14:13] [40,000/450,000] Checkpoint saved: /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/exp20-parallel-450k-v5/checkpoint_0040000.zip
[17:14:19]   Eval: gen_track=21.1r/148s ❌@148  mountain=15.7r/147s ❌@147
[17:14:20]   NEW BEST: 36.8 combined reward
------------------------------
| time/              |       |
|    fps             | 64    |
|    iterations      | 1     |
|    time_elapsed    | 63    |
|    total_timesteps | 45056 |
------------------------------
----------------------------------------
| time/                   |            |
|    fps                  | 37         |
|    iterations           | 2          |
|    time_elapsed         | 218        |
|    total_timesteps      | 49152      |
| train/                  |            |
|    approx_kl            | 0.74158007 |
|    clip_fraction        | 0.711      |
|    clip_range           | 0.2        |
|    entropy_loss         | -1.21      |
|    explained_variance   | 0.591      |
|    learning_rate        | 0.000725   |
|    loss                 | -0.0973    |
|    n_updates            | 110        |
|    policy_gradient_loss | -0.0698    |
|    std                  | 0.416      |
|    value_loss           | 0.462      |
----------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 33        |
|    iterations           | 3         |
|    time_elapsed         | 368       |
|    total_timesteps      | 53248     |
| train/                  |           |
|    approx_kl            | 0.8501822 |
|    clip_fraction        | 0.727     |
|    clip_range           | 0.2       |
|    entropy_loss         | -0.999    |
|    explained_variance   | 0.727     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.048    |
|    n_updates            | 120       |
|    policy_gradient_loss | -0.0694   |
|    std                  | 0.373     |
|    value_loss           | 0.399     |
---------------------------------------
----------------------------------------
| time/                   |            |
|    fps                  | 31         |
|    iterations           | 4          |
|    time_elapsed         | 523        |
|    total_timesteps      | 57344      |
| train/                  |            |
|    approx_kl            | 0.99027634 |
|    clip_fraction        | 0.729      |
|    clip_range           | 0.2        |
|    entropy_loss         | -0.791     |
|    explained_variance   | 0.804      |
|    learning_rate        | 0.000725   |
|    loss                 | -0.0752    |
|    n_updates            | 130        |
|    policy_gradient_loss | -0.0647    |
|    std                  | 0.335      |
|    value_loss           | 0.333      |
----------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 30        |
|    iterations           | 5         |
|    time_elapsed         | 669       |
|    total_timesteps      | 61440     |
| train/                  |           |
|    approx_kl            | 1.0802925 |
|    clip_fraction        | 0.727     |
|    clip_range           | 0.2       |
|    entropy_loss         | -0.588    |
|    explained_variance   | 0.81      |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0711   |
|    n_updates            | 140       |
|    policy_gradient_loss | -0.0571   |
|    std                  | 0.303     |
|    value_loss           | 0.459     |
---------------------------------------
[17:27:06] [60,000/450,000] Checkpoint saved: /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/exp20-parallel-450k-v5/checkpoint_0060000.zip
[17:27:13]   Eval: gen_track=6.6r/65s ❌@65  mountain=8.8r/112s ❌@112
------------------------------
| time/              |       |
|    fps             | 66    |
|    iterations      | 1     |
|    time_elapsed    | 61    |
|    total_timesteps | 65536 |
------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 39        |
|    iterations           | 2         |
|    time_elapsed         | 208       |
|    total_timesteps      | 69632     |
| train/                  |           |
|    approx_kl            | 1.1047575 |
|    clip_fraction        | 0.752     |
|    clip_range           | 0.2       |
|    entropy_loss         | -0.185    |
|    explained_variance   | 0.811     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0207   |
|    n_updates            | 160       |
|    policy_gradient_loss | -0.0497   |
|    std                  | 0.247     |
|    value_loss           | 0.504     |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 34        |
|    iterations           | 3         |
|    time_elapsed         | 355       |
|    total_timesteps      | 73728     |
| train/                  |           |
|    approx_kl            | 1.0727906 |
|    clip_fraction        | 0.757     |
|    clip_range           | 0.2       |
|    entropy_loss         | 0.0196    |
|    explained_variance   | 0.82      |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0671   |
|    n_updates            | 170       |
|    policy_gradient_loss | -0.0498   |
|    std                  | 0.223     |
|    value_loss           | 0.409     |
---------------------------------------
----------------------------------------
| time/                   |            |
|    fps                  | 32         |
|    iterations           | 4          |
|    time_elapsed         | 504        |
|    total_timesteps      | 77824      |
| train/                  |            |
|    approx_kl            | 0.91813445 |
|    clip_fraction        | 0.738      |
|    clip_range           | 0.2        |
|    entropy_loss         | 0.23       |
|    explained_variance   | 0.889      |
|    learning_rate        | 0.000725   |
|    loss                 | -0.0627    |
|    n_updates            | 180        |
|    policy_gradient_loss | -0.0518    |
|    std                  | 0.201      |
|    value_loss           | 0.324      |
----------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 30        |
|    iterations           | 5         |
|    time_elapsed         | 670       |
|    total_timesteps      | 81920     |
| train/                  |           |
|    approx_kl            | 1.0347024 |
|    clip_fraction        | 0.742     |
|    clip_range           | 0.2       |
|    entropy_loss         | 0.431     |
|    explained_variance   | 0.841     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.043    |
|    n_updates            | 190       |
|    policy_gradient_loss | -0.0455   |
|    std                  | 0.182     |
|    value_loss           | 0.473     |
---------------------------------------
[17:39:48] [80,000/450,000] Checkpoint saved: /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/exp20-parallel-450k-v5/checkpoint_0080000.zip
[17:39:55]   Eval: gen_track=37.2r/209s ❌@209  mountain=28.0r/208s ❌@208
[17:39:55]   NEW BEST: 65.2 combined reward
------------------------------
| time/              |       |
|    fps             | 70    |
|    iterations      | 1     |
|    time_elapsed    | 57    |
|    total_timesteps | 86016 |
------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 41        |
|    iterations           | 2         |
|    time_elapsed         | 197       |
|    total_timesteps      | 90112     |
| train/                  |           |
|    approx_kl            | 1.3096654 |
|    clip_fraction        | 0.777     |
|    clip_range           | 0.2       |
|    entropy_loss         | 0.824     |
|    explained_variance   | 0.676     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0305   |
|    n_updates            | 210       |
|    policy_gradient_loss | -0.0378   |
|    std                  | 0.149     |
|    value_loss           | 0.656     |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 36        |
|    iterations           | 3         |
|    time_elapsed         | 335       |
|    total_timesteps      | 94208     |
| train/                  |           |
|    approx_kl            | 1.4713185 |
|    clip_fraction        | 0.773     |
|    clip_range           | 0.2       |
|    entropy_loss         | 1.02      |
|    explained_variance   | 0.733     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0886   |
|    n_updates            | 220       |
|    policy_gradient_loss | -0.0435   |
|    std                  | 0.135     |
|    value_loss           | 0.46      |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 33        |
|    iterations           | 4         |
|    time_elapsed         | 482       |
|    total_timesteps      | 98304     |
| train/                  |           |
|    approx_kl            | 1.4152278 |
|    clip_fraction        | 0.768     |
|    clip_range           | 0.2       |
|    entropy_loss         | 1.22      |
|    explained_variance   | 0.816     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.045    |
|    n_updates            | 230       |
|    policy_gradient_loss | -0.0434   |
|    std                  | 0.122     |
|    value_loss           | 0.423     |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 32        |
|    iterations           | 5         |
|    time_elapsed         | 627       |
|    total_timesteps      | 102400    |
| train/                  |           |
|    approx_kl            | 1.4566097 |
|    clip_fraction        | 0.774     |
|    clip_range           | 0.2       |
|    entropy_loss         | 1.4       |
|    explained_variance   | 0.833     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0272   |
|    n_updates            | 240       |
|    policy_gradient_loss | -0.0242   |
|    std                  | 0.111     |
|    value_loss           | 0.603     |
---------------------------------------
[17:51:51] [100,000/450,000] Checkpoint saved: /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/exp20-parallel-450k-v5/checkpoint_0100000.zip
[17:51:58]   Eval: gen_track=28.5r/175s ❌@175  mountain=27.3r/175s ❌@175
-------------------------------
| time/              |        |
|    fps             | 80     |
|    iterations      | 1      |
|    time_elapsed    | 50     |
|    total_timesteps | 106496 |
-------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 42        |
|    iterations           | 2         |
|    time_elapsed         | 194       |
|    total_timesteps      | 110592    |
| train/                  |           |
|    approx_kl            | 1.3409375 |
|    clip_fraction        | 0.781     |
|    clip_range           | 0.2       |
|    entropy_loss         | 1.79      |
|    explained_variance   | 0.893     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0767   |
|    n_updates            | 260       |
|    policy_gradient_loss | -0.0304   |
|    std                  | 0.092     |
|    value_loss           | 0.312     |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 36        |
|    iterations           | 3         |
|    time_elapsed         | 340       |
|    total_timesteps      | 114688    |
| train/                  |           |
|    approx_kl            | 1.1949866 |
|    clip_fraction        | 0.775     |
|    clip_range           | 0.2       |
|    entropy_loss         | 1.95      |
|    explained_variance   | 0.851     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0391   |
|    n_updates            | 270       |
|    policy_gradient_loss | -0.0179   |
|    std                  | 0.0843    |
|    value_loss           | 0.466     |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 34        |
|    iterations           | 4         |
|    time_elapsed         | 479       |
|    total_timesteps      | 118784    |
| train/                  |           |
|    approx_kl            | 1.3683267 |
|    clip_fraction        | 0.795     |
|    clip_range           | 0.2       |
|    entropy_loss         | 2.11      |
|    explained_variance   | 0.847     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.04     |
|    n_updates            | 280       |
|    policy_gradient_loss | 0.0118    |
|    std                  | 0.0781    |
|    value_loss           | 0.518     |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 32        |
|    iterations           | 5         |
|    time_elapsed         | 635       |
|    total_timesteps      | 122880    |
| train/                  |           |
|    approx_kl            | 1.6239161 |
|    clip_fraction        | 0.793     |
|    clip_range           | 0.2       |
|    entropy_loss         | 2.25      |
|    explained_variance   | 0.839     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0657   |
|    n_updates            | 290       |
|    policy_gradient_loss | 0.007     |
|    std                  | 0.0734    |
|    value_loss           | 0.621     |
---------------------------------------
[18:04:01] [120,000/450,000] Checkpoint saved: /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/exp20-parallel-450k-v5/checkpoint_0120000.zip
[18:04:08]   Eval: gen_track=27.7r/180s ❌@180  mountain=25.7r/180s ❌@180
-------------------------------
| time/              |        |
|    fps             | 56     |
|    iterations      | 1      |
|    time_elapsed    | 73     |
|    total_timesteps | 126976 |
-------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 27        |
|    iterations           | 2         |
|    time_elapsed         | 292       |
|    total_timesteps      | 131072    |
| train/                  |           |
|    approx_kl            | 1.5144145 |
|    clip_fraction        | 0.816     |
|    clip_range           | 0.2       |
|    entropy_loss         | 2.45      |
|    explained_variance   | 0.721     |
|    learning_rate        | 0.000725  |
|    loss                 | 0.0266    |
|    n_updates            | 310       |
|    policy_gradient_loss | 0.0224    |
|    std                  | 0.0652    |
|    value_loss           | 0.869     |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 24        |
|    iterations           | 3         |
|    time_elapsed         | 502       |
|    total_timesteps      | 135168    |
| train/                  |           |
|    approx_kl            | 1.7737575 |
|    clip_fraction        | 0.817     |
|    clip_range           | 0.2       |
|    entropy_loss         | 2.59      |
|    explained_variance   | 0.751     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0113   |
|    n_updates            | 320       |
|    policy_gradient_loss | 0.0307    |
|    std                  | 0.061     |
|    value_loss           | 0.81      |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 22        |
|    iterations           | 4         |
|    time_elapsed         | 725       |
|    total_timesteps      | 139264    |
| train/                  |           |
|    approx_kl            | 2.3732958 |
|    clip_fraction        | 0.831     |
|    clip_range           | 0.2       |
|    entropy_loss         | 2.73      |
|    explained_variance   | 0.711     |
|    learning_rate        | 0.000725  |
|    loss                 | 0.0475    |
|    n_updates            | 330       |
|    policy_gradient_loss | 0.0432    |
|    std                  | 0.0575    |
|    value_loss           | 0.965     |
---------------------------------------
--------------------------------------
| time/                   |          |
|    fps                  | 21       |
|    iterations           | 5        |
|    time_elapsed         | 946      |
|    total_timesteps      | 143360   |
| train/                  |          |
|    approx_kl            | 3.434102 |
|    clip_fraction        | 0.86     |
|    clip_range           | 0.2      |
|    entropy_loss         | 2.83     |
|    explained_variance   | 0.574    |
|    learning_rate        | 0.000725 |
|    loss                 | 0.119    |
|    n_updates            | 340      |
|    policy_gradient_loss | 0.0767   |
|    std                  | 0.0555   |
|    value_loss           | 1.35     |
--------------------------------------
[18:22:13] [140,000/450,000] Checkpoint saved: /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/exp20-parallel-450k-v5/checkpoint_0140000.zip
[18:22:19]   Eval: gen_track=8.2r/79s ❌@79  mountain=8.9r/79s ❌@79
-------------------------------
| time/              |        |
|    fps             | 46     |
|    iterations      | 1      |
|    time_elapsed    | 88     |
|    total_timesteps | 147456 |
-------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 22        |
|    iterations           | 2         |
|    time_elapsed         | 357       |
|    total_timesteps      | 151552    |
| train/                  |           |
|    approx_kl            | 2.1389558 |
|    clip_fraction        | 0.838     |
|    clip_range           | 0.2       |
|    entropy_loss         | 2.94      |
|    explained_variance   | 0.596     |
|    learning_rate        | 0.000725  |
|    loss                 | 0.156     |
|    n_updates            | 360       |
|    policy_gradient_loss | 0.0675    |
|    std                  | 0.0521    |
|    value_loss           | 1.21      |
---------------------------------------
--------------------------------------
| time/                   |          |
|    fps                  | 19       |
|    iterations           | 3        |
|    time_elapsed         | 629      |
|    total_timesteps      | 155648   |
| train/                  |          |
|    approx_kl            | 4.391129 |
|    clip_fraction        | 0.863    |
|    clip_range           | 0.2      |
|    entropy_loss         | 3.04     |
|    explained_variance   | 0.656    |
|    learning_rate        | 0.000725 |
|    loss                 | 0.0562   |
|    n_updates            | 370      |
|    policy_gradient_loss | 0.0623   |
|    std                  | 0.0491   |
|    value_loss           | 0.813    |
--------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 18        |
|    iterations           | 4         |
|    time_elapsed         | 890       |
|    total_timesteps      | 159744    |
| train/                  |           |
|    approx_kl            | 2.7348022 |
|    clip_fraction        | 0.855     |
|    clip_range           | 0.2       |
|    entropy_loss         | 3.12      |
|    explained_variance   | 0.698     |
|    learning_rate        | 0.000725  |
|    loss                 | 0.0226    |
|    n_updates            | 380       |
|    policy_gradient_loss | 0.038     |
|    std                  | 0.047     |
|    value_loss           | 0.562     |
---------------------------------------
--------------------------------------
| time/                   |          |
|    fps                  | 18       |
|    iterations           | 5        |
|    time_elapsed         | 1133     |
|    total_timesteps      | 163840   |
| train/                  |          |
|    approx_kl            | 2.956015 |
|    clip_fraction        | 0.826    |
|    clip_range           | 0.2      |
|    entropy_loss         | 3.25     |
|    explained_variance   | 0.718    |
|    learning_rate        | 0.000725 |
|    loss                 | -0.0102  |
|    n_updates            | 390      |
|    policy_gradient_loss | 0.0226   |
|    std                  | 0.0448   |
|    value_loss           | 0.436    |
--------------------------------------
[18:43:43] [160,000/450,000] Checkpoint saved: /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/exp20-parallel-450k-v5/checkpoint_0160000.zip
[18:43:49]   Eval: gen_track=5.5r/69s ❌@69  mountain=7.4r/97s ❌@97
-------------------------------
| time/              |        |
|    fps             | 41     |
|    iterations      | 1      |
|    time_elapsed    | 98     |
|    total_timesteps | 167936 |
-------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 23        |
|    iterations           | 2         |
|    time_elapsed         | 350       |
|    total_timesteps      | 172032    |
| train/                  |           |
|    approx_kl            | 2.9826787 |
|    clip_fraction        | 0.813     |
|    clip_range           | 0.2       |
|    entropy_loss         | 3.5       |
|    explained_variance   | 0.778     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0133   |
|    n_updates            | 410       |
|    policy_gradient_loss | 0.000808  |
|    std                  | 0.0392    |
|    value_loss           | 0.266     |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 20        |
|    iterations           | 3         |
|    time_elapsed         | 601       |
|    total_timesteps      | 176128    |
| train/                  |           |
|    approx_kl            | 3.4679208 |
|    clip_fraction        | 0.801     |
|    clip_range           | 0.2       |
|    entropy_loss         | 3.68      |
|    explained_variance   | 0.786     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0178   |
|    n_updates            | 420       |
|    policy_gradient_loss | -0.0115   |
|    std                  | 0.0359    |
|    value_loss           | 0.278     |
---------------------------------------
--------------------------------------
| time/                   |          |
|    fps                  | 19       |
|    iterations           | 4        |
|    time_elapsed         | 842      |
|    total_timesteps      | 180224   |
| train/                  |          |
|    approx_kl            | 3.401013 |
|    clip_fraction        | 0.81     |
|    clip_range           | 0.2      |
|    entropy_loss         | 3.84     |
|    explained_variance   | 0.778    |
|    learning_rate        | 0.000725 |
|    loss                 | -0.0331  |
|    n_updates            | 430      |
|    policy_gradient_loss | -0.00564 |
|    std                  | 0.0333   |
|    value_loss           | 0.299    |
--------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 19        |
|    iterations           | 5         |
|    time_elapsed         | 1033      |
|    total_timesteps      | 184320    |
| train/                  |           |
|    approx_kl            | 2.4161968 |
|    clip_fraction        | 0.822     |
|    clip_range           | 0.2       |
|    entropy_loss         | 3.98      |
|    explained_variance   | 0.806     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0604   |
|    n_updates            | 440       |
|    policy_gradient_loss | 0.00771   |
|    std                  | 0.031     |
|    value_loss           | 0.283     |
---------------------------------------
[19:02:39] [180,000/450,000] Checkpoint saved: /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/exp20-parallel-450k-v5/checkpoint_0180000.zip
[19:02:45]   Eval: gen_track=6.7r/80s ❌@80  mountain=8.2r/97s ❌@97
-------------------------------
| time/              |        |
|    fps             | 45     |
|    iterations      | 1      |
|    time_elapsed    | 90     |
|    total_timesteps | 188416 |
-------------------------------
--------------------------------------
| time/                   |          |
|    fps                  | 30       |
|    iterations           | 2        |
|    time_elapsed         | 270      |
|    total_timesteps      | 192512   |
| train/                  |          |
|    approx_kl            | 4.115147 |
|    clip_fraction        | 0.836    |
|    clip_range           | 0.2      |
|    entropy_loss         | 4.22     |
|    explained_variance   | 0.78     |
|    learning_rate        | 0.000725 |
|    loss                 | 0.0814   |
|    n_updates            | 460      |
|    policy_gradient_loss | 0.0216   |
|    std                  | 0.0278   |
|    value_loss           | 0.338    |
--------------------------------------
--------------------------------------
| time/                   |          |
|    fps                  | 26       |
|    iterations           | 3        |
|    time_elapsed         | 464      |
|    total_timesteps      | 196608   |
| train/                  |          |
|    approx_kl            | 4.002183 |
|    clip_fraction        | 0.841    |
|    clip_range           | 0.2      |
|    entropy_loss         | 4.31     |
|    explained_variance   | 0.78     |
|    learning_rate        | 0.000725 |
|    loss                 | -0.0155  |
|    n_updates            | 470      |
|    policy_gradient_loss | 0.0366   |
|    std                  | 0.0268   |
|    value_loss           | 0.306    |
--------------------------------------
--------------------------------------
| time/                   |          |
|    fps                  | 25       |
|    iterations           | 4        |
|    time_elapsed         | 652      |
|    total_timesteps      | 200704   |
| train/                  |          |
|    approx_kl            | 3.378588 |
|    clip_fraction        | 0.822    |
|    clip_range           | 0.2      |
|    entropy_loss         | 4.41     |
|    explained_variance   | 0.807    |
|    learning_rate        | 0.000725 |
|    loss                 | -0.0678  |
|    n_updates            | 480      |
|    policy_gradient_loss | 0.0144   |
|    std                  | 0.0254   |
|    value_loss           | 0.339    |
--------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 24        |
|    iterations           | 5         |
|    time_elapsed         | 840       |
|    total_timesteps      | 204800    |
| train/                  |           |
|    approx_kl            | 3.6989145 |
|    clip_fraction        | 0.809     |
|    clip_range           | 0.2       |
|    entropy_loss         | 4.5       |
|    explained_variance   | 0.776     |
|    learning_rate        | 0.000725  |
|    loss                 | 0.0385    |
|    n_updates            | 490       |
|    policy_gradient_loss | 0.0184    |
|    std                  | 0.0242    |
|    value_loss           | 0.352     |
---------------------------------------
[19:18:17] [200,000/450,000] Checkpoint saved: /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/exp20-parallel-450k-v5/checkpoint_0200000.zip
[19:18:23]   Eval: gen_track=5.7r/69s ❌@69  mountain=6.7r/69s ❌@69
-------------------------------
| time/              |        |
|    fps             | 43     |
|    iterations      | 1      |
|    time_elapsed    | 94     |
|    total_timesteps | 208896 |
-------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 28        |
|    iterations           | 2         |
|    time_elapsed         | 287       |
|    total_timesteps      | 212992    |
| train/                  |           |
|    approx_kl            | 3.2493048 |
|    clip_fraction        | 0.827     |
|    clip_range           | 0.2       |
|    entropy_loss         | 4.71      |
|    explained_variance   | 0.795     |
|    learning_rate        | 0.000725  |
|    loss                 | 0.0462    |
|    n_updates            | 510       |
|    policy_gradient_loss | 0.03      |
|    std                  | 0.0219    |
|    value_loss           | 0.373     |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 25        |
|    iterations           | 3         |
|    time_elapsed         | 477       |
|    total_timesteps      | 217088    |
| train/                  |           |
|    approx_kl            | 4.8058205 |
|    clip_fraction        | 0.862     |
|    clip_range           | 0.2       |
|    entropy_loss         | 4.78      |
|    explained_variance   | 0.826     |
|    learning_rate        | 0.000725  |
|    loss                 | 0.0104    |
|    n_updates            | 520       |
|    policy_gradient_loss | 0.0622    |
|    std                  | 0.0211    |
|    value_loss           | 0.393     |
---------------------------------------
--------------------------------------
| time/                   |          |
|    fps                  | 24       |
|    iterations           | 4        |
|    time_elapsed         | 664      |
|    total_timesteps      | 221184   |
| train/                  |          |
|    approx_kl            | 4.024245 |
|    clip_fraction        | 0.842    |
|    clip_range           | 0.2      |
|    entropy_loss         | 4.87     |
|    explained_variance   | 0.796    |
|    learning_rate        | 0.000725 |
|    loss                 | 0.126    |
|    n_updates            | 530      |
|    policy_gradient_loss | 0.0408   |
|    std                  | 0.0205   |
|    value_loss           | 0.394    |
--------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 24        |
|    iterations           | 5         |
|    time_elapsed         | 852       |
|    total_timesteps      | 225280    |
| train/                  |           |
|    approx_kl            | 5.5941677 |
|    clip_fraction        | 0.824     |
|    clip_range           | 0.2       |
|    entropy_loss         | 4.95      |
|    explained_variance   | 0.799     |
|    learning_rate        | 0.000725  |
|    loss                 | 0.0566    |
|    n_updates            | 540       |
|    policy_gradient_loss | 0.13      |
|    std                  | 0.0194    |
|    value_loss           | 0.351     |
---------------------------------------
[19:34:05] [220,000/450,000] Checkpoint saved: /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/exp20-parallel-450k-v5/checkpoint_0220000.zip
[19:34:10]   Eval: gen_track=6.9r/82s ❌@82  mountain=7.9r/82s ❌@82
-------------------------------
| time/              |        |
|    fps             | 46     |
|    iterations      | 1      |
|    time_elapsed    | 88     |
|    total_timesteps | 229376 |
-------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 30        |
|    iterations           | 2         |
|    time_elapsed         | 266       |
|    total_timesteps      | 233472    |
| train/                  |           |
|    approx_kl            | 3.5075796 |
|    clip_fraction        | 0.857     |
|    clip_range           | 0.2       |
|    entropy_loss         | 5.08      |
|    explained_variance   | 0.834     |
|    learning_rate        | 0.000725  |
|    loss                 | 0.0347    |
|    n_updates            | 560       |
|    policy_gradient_loss | 0.0538    |
|    std                  | 0.0184    |
|    value_loss           | 0.273     |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 27        |
|    iterations           | 3         |
|    time_elapsed         | 454       |
|    total_timesteps      | 237568    |
| train/                  |           |
|    approx_kl            | 3.8775434 |
|    clip_fraction        | 0.863     |
|    clip_range           | 0.2       |
|    entropy_loss         | 5.15      |
|    explained_variance   | 0.875     |
|    learning_rate        | 0.000725  |
|    loss                 | -0.0319   |
|    n_updates            | 570       |
|    policy_gradient_loss | 0.0654    |
|    std                  | 0.0179    |
|    value_loss           | 0.223     |
---------------------------------------
--------------------------------------
| time/                   |          |
|    fps                  | 25       |
|    iterations           | 4        |
|    time_elapsed         | 633      |
|    total_timesteps      | 241664   |
| train/                  |          |
|    approx_kl            | 9.942043 |
|    clip_fraction        | 0.846    |
|    clip_range           | 0.2      |
|    entropy_loss         | 5.23     |
|    explained_variance   | 0.846    |
|    learning_rate        | 0.000725 |
|    loss                 | 0.0481   |
|    n_updates            | 580      |
|    policy_gradient_loss | 0.0485   |
|    std                  | 0.0172   |
|    value_loss           | 0.342    |
--------------------------------------
--------------------------------------
| time/                   |          |
|    fps                  | 25       |
|    iterations           | 5        |
|    time_elapsed         | 812      |
|    total_timesteps      | 245760   |
| train/                  |          |
|    approx_kl            | 6.776287 |
|    clip_fraction        | 0.889    |
|    clip_range           | 0.2      |
|    entropy_loss         | 5.23     |
|    explained_variance   | 0.857    |
|    learning_rate        | 0.000725 |
|    loss                 | 0.118    |
|    n_updates            | 590      |
|    policy_gradient_loss | 0.0907   |
|    std                  | 0.0175   |
|    value_loss           | 0.325    |
--------------------------------------
[19:49:11] [240,000/450,000] Checkpoint saved: /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/exp20-parallel-450k-v5/checkpoint_0240000.zip
[19:49:17]   Eval: gen_track=6.5r/80s ❌@80  mountain=8.2r/80s ❌@80
-------------------------------
| time/              |        |
|    fps             | 44     |
|    iterations      | 1      |
|    time_elapsed    | 91     |
|    total_timesteps | 249856 |
-------------------------------
--------------------------------------
| time/                   |          |
|    fps                  | 30       |
|    iterations           | 2        |
|    time_elapsed         | 271      |
|    total_timesteps      | 253952   |
| train/                  |          |
|    approx_kl            | 8.686437 |
|    clip_fraction        | 0.846    |
|    clip_range           | 0.2      |
|    entropy_loss         | 5.32     |
|    explained_variance   | 0.732    |
|    learning_rate        | 0.000725 |
|    loss                 | 0.169    |
|    n_updates            | 610      |
|    policy_gradient_loss | 0.0525   |
|    std                  | 0.016    |
|    value_loss           | 0.457    |
--------------------------------------
--------------------------------------
| time/                   |          |
|    fps                  | 27       |
|    iterations           | 3        |
|    time_elapsed         | 449      |
|    total_timesteps      | 258048   |
| train/                  |          |
|    approx_kl            | 8.093328 |
|    clip_fraction        | 0.853    |
|    clip_range           | 0.2      |
|    entropy_loss         | 5.4      |
|    explained_variance   | 0.811    |
|    learning_rate        | 0.000725 |
|    loss                 | 0.109    |
|    n_updates            | 620      |
|    policy_gradient_loss | 0.0624   |
|    std                  | 0.0155   |
|    value_loss           | 0.405    |
--------------------------------------
--------------------------------------
| time/                   |          |
|    fps                  | 25       |
|    iterations           | 4        |
|    time_elapsed         | 634      |
|    total_timesteps      | 262144   |
| train/                  |          |
|    approx_kl            | 8.560518 |
|    clip_fraction        | 0.825    |
|    clip_range           | 0.2      |
|    entropy_loss         | 5.47     |
|    explained_variance   | 0.801    |
|    learning_rate        | 0.000725 |
|    loss                 | 0.00362  |
|    n_updates            | 630      |
|    policy_gradient_loss | 0.0468   |
|    std                  | 0.0149   |
|    value_loss           | 0.417    |
--------------------------------------
--------------------------------------
| time/                   |          |
|    fps                  | 23       |
|    iterations           | 5        |
|    time_elapsed         | 890      |
|    total_timesteps      | 266240   |
| train/                  |          |
|    approx_kl            | 6.063898 |
|    clip_fraction        | 0.86     |
|    clip_range           | 0.2      |
|    entropy_loss         | 5.56     |
|    explained_variance   | 0.768    |
|    learning_rate        | 0.000725 |
|    loss                 | 0.00797  |
|    n_updates            | 640      |
|    policy_gradient_loss | 0.0776   |
|    std                  | 0.0146   |
|    value_loss           | 0.486    |
--------------------------------------
[20:06:13] [260,000/450,000] Checkpoint saved: /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/exp20-parallel-450k-v5/checkpoint_0260000.zip
[20:06:19]   Eval: gen_track=3.6r/59s ❌@59  mountain=6.4r/99s ❌@99
-------------------------------
| time/              |        |
|    fps             | 37     |
|    iterations      | 1      |
|    time_elapsed    | 110    |
|    total_timesteps | 270336 |
-------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 23        |
|    iterations           | 2         |
|    time_elapsed         | 343       |
|    total_timesteps      | 274432    |
| train/                  |           |
|    approx_kl            | 11.335977 |
|    clip_fraction        | 0.891     |
|    clip_range           | 0.2       |
|    entropy_loss         | 5.69      |
|    explained_variance   | 0.825     |
|    learning_rate        | 0.000725  |
|    loss                 | 0.0424    |
|    n_updates            | 660       |
|    policy_gradient_loss | 0.0787    |
|    std                  | 0.0136    |
|    value_loss           | 0.223     |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 20        |
|    iterations           | 3         |
|    time_elapsed         | 594       |
|    total_timesteps      | 278528    |
| train/                  |           |
|    approx_kl            | 11.321661 |
|    clip_fraction        | 0.852     |
|    clip_range           | 0.2       |
|    entropy_loss         | 5.78      |
|    explained_variance   | 0.808     |
|    learning_rate        | 0.000725  |
|    loss                 | 0.104     |
|    n_updates            | 670       |
|    policy_gradient_loss | 0.0575    |
|    std                  | 0.0128    |
|    value_loss           | 0.213     |
---------------------------------------
---------------------------------------
| time/                   |           |
|    fps                  | 19        |
|    iterations           | 4         |
|    time_elapsed         | 860       |
|    total_timesteps      | 282624    |
| train/                  |           |
|    approx_kl            | 18.683723 |
|    clip_fraction        | 0.862     |
|    clip_range           | 0.2       |
|    entropy_loss         | 5.93      |
|    explained_variance   | 0.844     |
|    learning_rate        | 0.000725  |
|    loss                 | 0.0861    |
|    n_updates            | 680       |
|    policy_gradient_loss | 0.076     |
|    std                  | 0.0119    |
|    value_loss           | 0.183     |
---------------------------------------