paulh
/
donkeycar-rl-autoresearch
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

wave3: autoresearch trial 20 results 

Agent: pi
Tests: N/A
Tests-Added: 0
TypeScript: N/A
This commit is contained in:
Paul Huliganga 2026-04-16 14:10:06 -04:00
parent 45b057e9c1
commit 96c49dd057
2 changed files with 116 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

111
agent/outerloop-results/autoresearch_phase4_log.txt
View File

@ -599,3 +599,114 @@
[2026-04-16 08:43:17] score=1435.04 params={'learning_rate': 0.0007252855740444645, 'steps_per_switch': 6851, 'total_timesteps': 89893}
[2026-04-16 08:43:17] score=176.67 params={'learning_rate': 0.0016223486895735558, 'steps_per_switch': 5524, 'total_timesteps': 79532}
[2026-04-16 08:43:17] score=141.35 params={'learning_rate': 0.0010578566658808422, 'steps_per_switch': 4587, 'total_timesteps': 89501}
[2026-04-16 08:43:19] [Wave4] ✅ Git push complete after trial 15
[2026-04-16 08:43:21]
[Wave4] ========== Trial 16/25 ==========
[2026-04-16 08:43:21] [Wave4] GP UCB top-5 proposals:
[2026-04-16 08:43:21] UCB=3.6132 mu=2.7065 σ=0.4533 params={'learning_rate': 0.00040295487555843073, 'steps_per_switch': 3490, 'total_timesteps': 72578}
[2026-04-16 08:43:21] UCB=3.6040 mu=2.6026 σ=0.5007 params={'learning_rate': 0.0004730969154346568, 'steps_per_switch': 3501, 'total_timesteps': 73673}
[2026-04-16 08:43:21] UCB=3.5949 mu=2.3139 σ=0.6405 params={'learning_rate': 0.0007493214757809492, 'steps_per_switch': 3197, 'total_timesteps': 74747}
[2026-04-16 08:43:21] UCB=3.3717 mu=2.1460 σ=0.6128 params={'learning_rate': 0.0008356081352627151, 'steps_per_switch': 3341, 'total_timesteps': 76184}
[2026-04-16 08:43:21] UCB=3.2598 mu=1.9831 σ=0.6384 params={'learning_rate': 0.00015673628608872705, 'steps_per_switch': 3392, 'total_timesteps': 63890}
[2026-04-16 08:43:21] [Wave4] Proposed params: {'learning_rate': 0.00040295487555843073, 'steps_per_switch': 3490, 'total_timesteps': 72578}
[2026-04-16 08:43:23] [Wave4] Launching trial 16: {'learning_rate': 0.00040295487555843073, 'steps_per_switch': 3490, 'total_timesteps': 72578}
[2026-04-16 08:43:23] [Wave4] Command: python3 /home/paulh/projects/donkeycar-rl-autoresearch/agent/multitrack_runner.py --total-timesteps 72578 --steps-per-switch 3490 --learning-rate 0.00040295487555843073 --eval-episodes 3 --save-dir /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0016
[2026-04-16 09:52:25] [Wave4] Trial 16 finished in 4142.2s, rc=0
[2026-04-16 09:52:25] [Wave4] Parsed: combined=59.9996 mini_monaco=59.9996
[2026-04-16 09:52:25] [Wave4] ===== Trial 16 Summary =====
[2026-04-16 09:52:25] GP data points : 16
[2026-04-16 09:52:25] Wave4 Champion: trial=3 score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 157743}
[2026-04-16 09:52:25] Top 5:
[2026-04-16 09:52:25] score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 90000}
[2026-04-16 09:52:25] score=1573.07 params={'learning_rate': 0.00033896311195395496, 'steps_per_switch': 5448, 'total_timesteps': 69359}
[2026-04-16 09:52:25] score=1435.04 params={'learning_rate': 0.0007252855740444645, 'steps_per_switch': 6851, 'total_timesteps': 89893}
[2026-04-16 09:52:25] score=176.67 params={'learning_rate': 0.0016223486895735558, 'steps_per_switch': 5524, 'total_timesteps': 79532}
[2026-04-16 09:52:25] score=141.35 params={'learning_rate': 0.0010578566658808422, 'steps_per_switch': 4587, 'total_timesteps': 89501}
[2026-04-16 09:52:27]
[Wave4] ========== Trial 17/25 ==========
[2026-04-16 09:52:27] [Wave4] GP UCB top-5 proposals:
[2026-04-16 09:52:27] UCB=4.4175 mu=3.0206 σ=0.6985 params={'learning_rate': 0.00011371306784012085, 'steps_per_switch': 4558, 'total_timesteps': 61409}
[2026-04-16 09:52:27] UCB=3.8442 mu=2.8720 σ=0.4861 params={'learning_rate': 0.00015521967796620274, 'steps_per_switch': 5888, 'total_timesteps': 64258}
[2026-04-16 09:52:27] UCB=3.8027 mu=2.6781 σ=0.5623 params={'learning_rate': 0.0001942087695905749, 'steps_per_switch': 3887, 'total_timesteps': 63994}
[2026-04-16 09:52:27] UCB=3.7728 mu=2.8641 σ=0.4544 params={'learning_rate': 0.00019296340532126325, 'steps_per_switch': 4356, 'total_timesteps': 65008}
[2026-04-16 09:52:27] UCB=3.7728 mu=2.4744 σ=0.6492 params={'learning_rate': 0.0004244302343635027, 'steps_per_switch': 4871, 'total_timesteps': 61656}
[2026-04-16 09:52:27] [Wave4] Proposed params: {'learning_rate': 0.00011371306784012085, 'steps_per_switch': 4558, 'total_timesteps': 61409}
[2026-04-16 09:52:29] [Wave4] Launching trial 17: {'learning_rate': 0.00011371306784012085, 'steps_per_switch': 4558, 'total_timesteps': 61409}
[2026-04-16 09:52:29] [Wave4] Command: python3 /home/paulh/projects/donkeycar-rl-autoresearch/agent/multitrack_runner.py --total-timesteps 61409 --steps-per-switch 4558 --learning-rate 0.00011371306784012085 --eval-episodes 3 --save-dir /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0017
[2026-04-16 10:54:36] [Wave4] Trial 17 finished in 3727.6s, rc=0
[2026-04-16 10:54:36] [Wave4] Parsed: combined=106.1213 mini_monaco=106.1213
[2026-04-16 10:54:36] [Wave4] ===== Trial 17 Summary =====
[2026-04-16 10:54:36] GP data points : 17
[2026-04-16 10:54:36] Wave4 Champion: trial=3 score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 157743}
[2026-04-16 10:54:36] Top 5:
[2026-04-16 10:54:36] score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 90000}
[2026-04-16 10:54:36] score=1573.07 params={'learning_rate': 0.00033896311195395496, 'steps_per_switch': 5448, 'total_timesteps': 69359}
[2026-04-16 10:54:36] score=1435.04 params={'learning_rate': 0.0007252855740444645, 'steps_per_switch': 6851, 'total_timesteps': 89893}
[2026-04-16 10:54:36] score=176.67 params={'learning_rate': 0.0016223486895735558, 'steps_per_switch': 5524, 'total_timesteps': 79532}
[2026-04-16 10:54:36] score=141.35 params={'learning_rate': 0.0010578566658808422, 'steps_per_switch': 4587, 'total_timesteps': 89501}
[2026-04-16 10:54:38]
[Wave4] ========== Trial 18/25 ==========
[2026-04-16 10:54:38] [Wave4] GP UCB top-5 proposals:
[2026-04-16 10:54:38] UCB=2.4022 mu=1.7625 σ=0.3199 params={'learning_rate': 0.00014739138140971148, 'steps_per_switch': 5692, 'total_timesteps': 73312}
[2026-04-16 10:54:38] UCB=2.1766 mu=1.8937 σ=0.1414 params={'learning_rate': 0.0003092758146459376, 'steps_per_switch': 5733, 'total_timesteps': 72147}
[2026-04-16 10:54:38] UCB=2.0283 mu=1.8693 σ=0.0795 params={'learning_rate': 0.00038035350750885823, 'steps_per_switch': 5423, 'total_timesteps': 70634}
[2026-04-16 10:54:38] UCB=2.0025 mu=0.0985 σ=0.9520 params={'learning_rate': 0.0019422789362836088, 'steps_per_switch': 9820, 'total_timesteps': 71591}
[2026-04-16 10:54:38] UCB=1.9979 mu=1.7539 σ=0.1220 params={'learning_rate': 0.0002958405723304147, 'steps_per_switch': 5704, 'total_timesteps': 67124}
[2026-04-16 10:54:38] [Wave4] Proposed params: {'learning_rate': 0.00014739138140971148, 'steps_per_switch': 5692, 'total_timesteps': 73312}
[2026-04-16 10:54:40] [Wave4] Launching trial 18: {'learning_rate': 0.00014739138140971148, 'steps_per_switch': 5692, 'total_timesteps': 73312}
[2026-04-16 10:54:40] [Wave4] Command: python3 /home/paulh/projects/donkeycar-rl-autoresearch/agent/multitrack_runner.py --total-timesteps 73312 --steps-per-switch 5692 --learning-rate 0.00014739138140971148 --eval-episodes 3 --save-dir /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0018
[2026-04-16 11:58:07] [Wave4] Trial 18 finished in 3806.9s, rc=0
[2026-04-16 11:58:07] [Wave4] Parsed: combined=115.6774 mini_monaco=115.6774
[2026-04-16 11:58:07] [Wave4] ===== Trial 18 Summary =====
[2026-04-16 11:58:07] GP data points : 18
[2026-04-16 11:58:07] Wave4 Champion: trial=3 score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 157743}
[2026-04-16 11:58:07] Top 5:
[2026-04-16 11:58:07] score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 90000}
[2026-04-16 11:58:07] score=1573.07 params={'learning_rate': 0.00033896311195395496, 'steps_per_switch': 5448, 'total_timesteps': 69359}
[2026-04-16 11:58:07] score=1435.04 params={'learning_rate': 0.0007252855740444645, 'steps_per_switch': 6851, 'total_timesteps': 89893}
[2026-04-16 11:58:07] score=176.67 params={'learning_rate': 0.0016223486895735558, 'steps_per_switch': 5524, 'total_timesteps': 79532}
[2026-04-16 11:58:07] score=141.35 params={'learning_rate': 0.0010578566658808422, 'steps_per_switch': 4587, 'total_timesteps': 89501}
[2026-04-16 11:58:09]
[Wave4] ========== Trial 19/25 ==========
[2026-04-16 11:58:09] [Wave4] GP UCB top-5 proposals:
[2026-04-16 11:58:09] UCB=4.2331 mu=3.1539 σ=0.5396 params={'learning_rate': 0.0006672844816013197, 'steps_per_switch': 4747, 'total_timesteps': 64179}
[2026-04-16 11:58:09] UCB=3.9367 mu=2.3643 σ=0.7862 params={'learning_rate': 0.000973356801504352, 'steps_per_switch': 4731, 'total_timesteps': 64636}
[2026-04-16 11:58:09] UCB=3.8093 mu=2.3419 σ=0.7337 params={'learning_rate': 0.0009048073949917447, 'steps_per_switch': 4137, 'total_timesteps': 65312}
[2026-04-16 11:58:09] UCB=3.7049 mu=2.1579 σ=0.7735 params={'learning_rate': 0.0008513107736880863, 'steps_per_switch': 6388, 'total_timesteps': 62234}
[2026-04-16 11:58:09] UCB=3.6513 mu=2.2696 σ=0.6908 params={'learning_rate': 0.0006744614059620855, 'steps_per_switch': 4569, 'total_timesteps': 60715}
[2026-04-16 11:58:09] [Wave4] Proposed params: {'learning_rate': 0.0006672844816013197, 'steps_per_switch': 4747, 'total_timesteps': 64179}
[2026-04-16 11:58:11] [Wave4] Launching trial 19: {'learning_rate': 0.0006672844816013197, 'steps_per_switch': 4747, 'total_timesteps': 64179}
[2026-04-16 11:58:11] [Wave4] Command: python3 /home/paulh/projects/donkeycar-rl-autoresearch/agent/multitrack_runner.py --total-timesteps 64179 --steps-per-switch 4747 --learning-rate 0.0006672844816013197 --eval-episodes 3 --save-dir /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0019
[2026-04-16 13:10:23] [Wave4] Trial 19 finished in 4331.6s, rc=0
[2026-04-16 13:10:23] [Wave4] Parsed: combined=230.9827 mini_monaco=230.9827
[2026-04-16 13:10:23] [Wave4] ===== Trial 19 Summary =====
[2026-04-16 13:10:23] GP data points : 19
[2026-04-16 13:10:23] Wave4 Champion: trial=3 score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 157743}
[2026-04-16 13:10:23] Top 5:
[2026-04-16 13:10:23] score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 90000}
[2026-04-16 13:10:23] score=1573.07 params={'learning_rate': 0.00033896311195395496, 'steps_per_switch': 5448, 'total_timesteps': 69359}
[2026-04-16 13:10:23] score=1435.04 params={'learning_rate': 0.0007252855740444645, 'steps_per_switch': 6851, 'total_timesteps': 89893}
[2026-04-16 13:10:23] score=230.98 params={'learning_rate': 0.0006672844816013197, 'steps_per_switch': 4747, 'total_timesteps': 64179}
[2026-04-16 13:10:23] score=176.67 params={'learning_rate': 0.0016223486895735558, 'steps_per_switch': 5524, 'total_timesteps': 79532}
[2026-04-16 13:10:25]
[Wave4] ========== Trial 20/25 ==========
[2026-04-16 13:10:25] [Wave4] GP UCB top-5 proposals:
[2026-04-16 13:10:25] UCB=2.2341 mu=1.6242 σ=0.3049 params={'learning_rate': 0.0007104769870597334, 'steps_per_switch': 5822, 'total_timesteps': 73160}
[2026-04-16 13:10:25] UCB=2.2038 mu=1.8331 σ=0.1853 params={'learning_rate': 0.000688892492740156, 'steps_per_switch': 5559, 'total_timesteps': 78911}
[2026-04-16 13:10:25] UCB=2.2000 mu=1.6830 σ=0.2585 params={'learning_rate': 0.0006312216002702526, 'steps_per_switch': 7125, 'total_timesteps': 80150}
[2026-04-16 13:10:25] UCB=2.1950 mu=1.5842 σ=0.3054 params={'learning_rate': 0.0009544613332763077, 'steps_per_switch': 3097, 'total_timesteps': 89206}
[2026-04-16 13:10:25] UCB=2.0924 mu=1.5207 σ=0.2859 params={'learning_rate': 0.0006628882948205277, 'steps_per_switch': 6497, 'total_timesteps': 73747}
[2026-04-16 13:10:25] [Wave4] Proposed params: {'learning_rate': 0.0007104769870597334, 'steps_per_switch': 5822, 'total_timesteps': 73160}
[2026-04-16 13:10:27] [Wave4] Launching trial 20: {'learning_rate': 0.0007104769870597334, 'steps_per_switch': 5822, 'total_timesteps': 73160}
[2026-04-16 13:10:27] [Wave4] Command: python3 /home/paulh/projects/donkeycar-rl-autoresearch/agent/multitrack_runner.py --total-timesteps 73160 --steps-per-switch 5822 --learning-rate 0.0007104769870597334 --eval-episodes 3 --save-dir /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0020
[2026-04-16 14:10:06] [Wave4] Trial 20 finished in 3578.9s, rc=0
[2026-04-16 14:10:06] [Wave4] Parsed: combined=20.9587 mini_monaco=20.9587
[2026-04-16 14:10:06] [Wave4] ===== Trial 20 Summary =====
[2026-04-16 14:10:06] GP data points : 20
[2026-04-16 14:10:06] Wave4 Champion: trial=3 score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 157743}
[2026-04-16 14:10:06] Top 5:
[2026-04-16 14:10:06] score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 90000}
[2026-04-16 14:10:06] score=1573.07 params={'learning_rate': 0.00033896311195395496, 'steps_per_switch': 5448, 'total_timesteps': 69359}
[2026-04-16 14:10:06] score=1435.04 params={'learning_rate': 0.0007252855740444645, 'steps_per_switch': 6851, 'total_timesteps': 89893}
[2026-04-16 14:10:06] score=230.98 params={'learning_rate': 0.0006672844816013197, 'steps_per_switch': 4747, 'total_timesteps': 64179}
[2026-04-16 14:10:06] score=176.67 params={'learning_rate': 0.0016223486895735558, 'steps_per_switch': 5524, 'total_timesteps': 79532}

5
agent/outerloop-results/autoresearch_results_phase4.jsonl
View File

@ -13,3 +13,8 @@
{"trial": 13, "timestamp": "2026-04-16T07:01:28.371417", "params": {"learning_rate": 0.0019124682919611855, "steps_per_switch": 3574, "total_timesteps": 73915}, "combined_test_score": 86.8189, "mini_monaco_reward": 86.8189, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0013/model.zip", "champion": false, "run_status": "ok", "elapsed_sec": 3605.6811096668243}
{"trial": 14, "timestamp": "2026-04-16T07:54:33.441210", "params": {"learning_rate": 0.00033896311195395496, "steps_per_switch": 5448, "total_timesteps": 69359}, "combined_test_score": 1573.0703, "mini_monaco_reward": 1573.0703, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0014/model.zip", "champion": false, "run_status": "ok", "elapsed_sec": 3181.0467441082}
{"trial": 15, "timestamp": "2026-04-16T08:43:17.443930", "params": {"learning_rate": 0.0003994526493374, "steps_per_switch": 7747, "total_timesteps": 70559}, "combined_test_score": 111.2516, "mini_monaco_reward": 111.2516, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0015/model.zip", "champion": false, "run_status": "ok", "elapsed_sec": 2919.980232000351}
{"trial": 16, "timestamp": "2026-04-16T09:52:25.241431", "params": {"learning_rate": 0.00040295487555843073, "steps_per_switch": 3490, "total_timesteps": 72578}, "combined_test_score": 59.9996, "mini_monaco_reward": 59.9996, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0016/model.zip", "champion": false, "run_status": "ok", "elapsed_sec": 4142.210912942886}
{"trial": 17, "timestamp": "2026-04-16T10:54:36.880476", "params": {"learning_rate": 0.00011371306784012085, "steps_per_switch": 4558, "total_timesteps": 61409}, "combined_test_score": 106.1213, "mini_monaco_reward": 106.1213, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0017/model.zip", "champion": false, "run_status": "ok", "elapsed_sec": 3727.5582659244537}
{"trial": 18, "timestamp": "2026-04-16T11:58:07.842111", "params": {"learning_rate": 0.00014739138140971148, "steps_per_switch": 5692, "total_timesteps": 73312}, "combined_test_score": 115.6774, "mini_monaco_reward": 115.6774, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0018/model.zip", "champion": false, "run_status": "ok", "elapsed_sec": 3806.9049928188324}
{"trial": 19, "timestamp": "2026-04-16T13:10:23.540177", "params": {"learning_rate": 0.0006672844816013197, "steps_per_switch": 4747, "total_timesteps": 64179}, "combined_test_score": 230.9827, "mini_monaco_reward": 230.9827, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0019/model.zip", "champion": false, "run_status": "ok", "elapsed_sec": 4331.6394629478455}
{"trial": 20, "timestamp": "2026-04-16T14:10:06.493399", "params": {"learning_rate": 0.0007104769870597334, "steps_per_switch": 5822, "total_timesteps": 73160}, "combined_test_score": 20.9587, "mini_monaco_reward": 20.9587, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0020/model.zip", "champion": false, "run_status": "ok", "elapsed_sec": 3578.8967142105103}