donkeycar-rl-autoresearch/agent/outerloop-results/autoresearch_phase4_log.txt

122 lines
12 KiB
Plaintext
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

[2026-04-14 22:40:44] =================================================================
[2026-04-14 22:40:44] [Wave4] Multi-Track Autoresearch — GP+UCB Generalization Search
[2026-04-14 22:40:44] [Wave4] Training tracks : generated_track, mountain_track (no generated_road, no warm-start)
[2026-04-14 22:40:44] [Wave4] Test tracks : mini_monaco only (zero-shot; warren removed — broken done condition)
[2026-04-14 22:40:44] [Wave4] Max trials : 25 | kappa=2.0 | push every 5
[2026-04-14 22:40:44] [Wave4] Results file : /home/paulh/projects/donkeycar-rl-autoresearch/agent/outerloop-results/autoresearch_results_phase4.jsonl
[2026-04-14 22:40:44] [Wave4] Champion dir : /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-champion
[2026-04-14 22:40:44] [Wave4] Warm start : NONE (training from scratch each trial)
[2026-04-14 22:40:44] =================================================================
[2026-04-14 22:40:44] [Wave4] Loaded 0 existing Phase 3 results.
[2026-04-14 22:40:44] [Wave4] No Wave 3 champion yet.
[2026-04-14 22:40:44] [Wave4] Starting from trial 1.
[2026-04-14 22:40:44]
[Wave4] ========== Trial 1/25 ==========
[2026-04-14 22:40:44] [Wave4] Seed trial 1/2: using hardcoded params.
[2026-04-14 22:40:44] [Wave4] Proposed params: {'learning_rate': 0.0003, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-14 22:40:46] [Wave4] Launching trial 1: {'learning_rate': 0.0003, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-14 22:40:46] [Wave4] Command: python3 /home/paulh/projects/donkeycar-rl-autoresearch/agent/multitrack_runner.py --total-timesteps 80000 --steps-per-switch 6000 --learning-rate 0.0003 --eval-episodes 3 --save-dir /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0001
[2026-04-14 22:44:24] =================================================================
[2026-04-14 22:44:24] [Wave4] Multi-Track Autoresearch — GP+UCB Generalization Search
[2026-04-14 22:44:24] [Wave4] Training tracks : generated_track, mountain_track (no generated_road, no warm-start)
[2026-04-14 22:44:24] [Wave4] Test tracks : mini_monaco only (zero-shot; warren removed — broken done condition)
[2026-04-14 22:44:24] [Wave4] Max trials : 25 | kappa=2.0 | push every 5
[2026-04-14 22:44:24] [Wave4] Results file : /home/paulh/projects/donkeycar-rl-autoresearch/agent/outerloop-results/autoresearch_results_phase4.jsonl
[2026-04-14 22:44:24] [Wave4] Champion dir : /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-champion
[2026-04-14 22:44:24] [Wave4] Warm start : NONE (training from scratch each trial)
[2026-04-14 22:44:24] =================================================================
[2026-04-14 22:44:24] [Wave4] Loaded 0 existing Phase 3 results.
[2026-04-14 22:44:24] [Wave4] No Wave 3 champion yet.
[2026-04-14 22:44:24] [Wave4] Starting from trial 1.
[2026-04-14 22:44:24]
[Wave4] ========== Trial 1/25 ==========
[2026-04-14 22:44:24] [Wave4] Seed trial 1/2: using hardcoded params.
[2026-04-14 22:44:24] [Wave4] Proposed params: {'learning_rate': 0.0003, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-14 22:44:26] [Wave4] Launching trial 1: {'learning_rate': 0.0003, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-14 22:44:26] [Wave4] Command: python3 /home/paulh/projects/donkeycar-rl-autoresearch/agent/multitrack_runner.py --total-timesteps 80000 --steps-per-switch 6000 --learning-rate 0.0003 --eval-episodes 3 --save-dir /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0001
[2026-04-15 00:02:45] [Wave4] Trial 1 finished in 4699.3s, rc=0
[2026-04-15 00:02:45] [Wave4] Parsed: combined=45.6693 mini_monaco=45.6693
[2026-04-15 00:02:45] [Champion] 🏆 NEW BEST! Trial 1: score=45.67 (mini_monaco=45.7) params={'learning_rate': 0.0003, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-15 00:02:45] [Wave4] ===== Trial 1 Summary =====
[2026-04-15 00:02:45] GP data points : 1
[2026-04-15 00:02:45] Wave4 Champion: trial=1 score=45.67 params={'learning_rate': 0.0003, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-15 00:02:45] Top 5:
[2026-04-15 00:02:45] score=45.67 params={'learning_rate': 0.0003, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-15 00:02:47]
[Wave4] ========== Trial 2/25 ==========
[2026-04-15 00:02:47] [Wave4] Seed trial 2/2: using hardcoded params.
[2026-04-15 00:02:47] [Wave4] Proposed params: {'learning_rate': 0.001, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-15 00:02:49] [Wave4] Launching trial 2: {'learning_rate': 0.001, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-15 00:02:49] [Wave4] Command: python3 /home/paulh/projects/donkeycar-rl-autoresearch/agent/multitrack_runner.py --total-timesteps 80000 --steps-per-switch 6000 --learning-rate 0.001 --eval-episodes 3 --save-dir /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0002
[2026-04-15 01:21:38] [Wave4] Trial 2 finished in 4728.4s, rc=0
[2026-04-15 01:21:38] [Wave4] Parsed: combined=222.0731 mini_monaco=222.0731
[2026-04-15 01:21:38] [Champion] 🏆 NEW BEST! Trial 2: score=222.07 (mini_monaco=222.1) params={'learning_rate': 0.001, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-15 01:21:38] [Wave4] ===== Trial 2 Summary =====
[2026-04-15 01:21:38] GP data points : 2
[2026-04-15 01:21:38] Wave4 Champion: trial=2 score=222.07 params={'learning_rate': 0.001, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-15 01:21:38] Top 5:
[2026-04-15 01:21:38] score=222.07 params={'learning_rate': 0.001, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-15 01:21:38] score=45.67 params={'learning_rate': 0.0003, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-15 01:21:40]
[Wave4] ========== Trial 3/25 ==========
[2026-04-15 01:21:40] [Wave4] Only 2 results — using random proposal.
[2026-04-15 01:21:40] [Wave4] Proposed params: {'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 157743}
[2026-04-15 01:21:42] [Wave4] Launching trial 3: {'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 157743}
[2026-04-15 01:21:42] [Wave4] Command: python3 /home/paulh/projects/donkeycar-rl-autoresearch/agent/multitrack_runner.py --total-timesteps 157743 --steps-per-switch 17499 --learning-rate 0.0006852550685205609 --eval-episodes 3 --save-dir /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0003
[2026-04-15 03:15:46] [Wave4] Trial 3 finished in 6843.7s, rc=0
[2026-04-15 03:15:46] [Wave4] Parsed: combined=1943.1038 mini_monaco=1943.1038
[2026-04-15 03:15:46] [Champion] 🏆 NEW BEST! Trial 3: score=1943.10 (mini_monaco=1943.1) params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 157743}
[2026-04-15 03:15:46] [Wave4] ===== Trial 3 Summary =====
[2026-04-15 03:15:46] GP data points : 3
[2026-04-15 03:15:46] Wave4 Champion: trial=3 score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 157743}
[2026-04-15 03:15:46] Top 5:
[2026-04-15 03:15:46] score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 157743}
[2026-04-15 03:15:46] score=222.07 params={'learning_rate': 0.001, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-15 03:15:46] score=45.67 params={'learning_rate': 0.0003, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-15 03:15:48]
[Wave4] ========== Trial 4/25 ==========
[2026-04-15 03:15:48] [Wave4] GP UCB top-5 proposals:
[2026-04-15 03:15:48] UCB=2.4560 mu=0.8788 σ=0.7886 params={'learning_rate': 0.0003250095463348546, 'steps_per_switch': 19054, 'total_timesteps': 197116}
[2026-04-15 03:15:48] UCB=2.4518 mu=0.8393 σ=0.8062 params={'learning_rate': 0.00121703003154963, 'steps_per_switch': 16951, 'total_timesteps': 180865}
[2026-04-15 03:15:48] UCB=2.4512 mu=0.7637 σ=0.8437 params={'learning_rate': 0.00036067077082995895, 'steps_per_switch': 16532, 'total_timesteps': 211219}
[2026-04-15 03:15:48] UCB=2.4501 mu=0.9283 σ=0.7609 params={'learning_rate': 0.0005325315186424085, 'steps_per_switch': 18992, 'total_timesteps': 205595}
[2026-04-15 03:15:48] UCB=2.4492 mu=0.9106 σ=0.7693 params={'learning_rate': 0.001163360064352729, 'steps_per_switch': 19652, 'total_timesteps': 151744}
[2026-04-15 03:15:48] [Wave4] Proposed params: {'learning_rate': 0.0003250095463348546, 'steps_per_switch': 19054, 'total_timesteps': 197116}
[2026-04-15 03:15:50] [Wave4] Launching trial 4: {'learning_rate': 0.0003250095463348546, 'steps_per_switch': 19054, 'total_timesteps': 197116}
[2026-04-15 03:15:50] [Wave4] Command: python3 /home/paulh/projects/donkeycar-rl-autoresearch/agent/multitrack_runner.py --total-timesteps 197116 --steps-per-switch 19054 --learning-rate 0.0003250095463348546 --eval-episodes 3 --save-dir /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0004
[2026-04-15 05:15:51] [Wave4] Trial 4 TIMED OUT — killing runner.
[2026-04-15 05:15:51] [Wave4] Trial 4 finished in 7200.5s, rc=-9
[2026-04-15 05:15:51] [Wave4] Parsed: combined=None mini_monaco=None
[2026-04-15 05:15:51] [Wave4] ⚠️ No test score parsed — defaulting to 0.0
[2026-04-15 05:15:51] [Wave4] combined_test_score=0 — excluded from GP (crash/timeout).
[2026-04-15 05:15:51] [Wave4] ===== Trial 4 Summary =====
[2026-04-15 05:15:51] GP data points : 3
[2026-04-15 05:15:51] Wave4 Champion: trial=3 score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 157743}
[2026-04-15 05:15:51] Top 5:
[2026-04-15 05:15:51] score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 157743}
[2026-04-15 05:15:51] score=222.07 params={'learning_rate': 0.001, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-15 05:15:51] score=45.67 params={'learning_rate': 0.0003, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-15 05:15:53]
[Wave4] ========== Trial 5/25 ==========
[2026-04-15 05:15:53] [Wave4] GP UCB top-5 proposals:
[2026-04-15 05:15:53] UCB=2.4597 mu=0.8329 σ=0.8134 params={'learning_rate': 0.0003927960467617446, 'steps_per_switch': 19892, 'total_timesteps': 201785}
[2026-04-15 05:15:53] UCB=2.4568 mu=0.8585 σ=0.7991 params={'learning_rate': 0.0011330710879806035, 'steps_per_switch': 18089, 'total_timesteps': 193054}
[2026-04-15 05:15:53] UCB=2.4560 mu=0.7832 σ=0.8364 params={'learning_rate': 0.0006110661120319741, 'steps_per_switch': 17141, 'total_timesteps': 219583}
[2026-04-15 05:15:53] UCB=2.4560 mu=0.8338 σ=0.8111 params={'learning_rate': 0.000602366907571214, 'steps_per_switch': 16527, 'total_timesteps': 215069}
[2026-04-15 05:15:53] UCB=2.4522 mu=0.8120 σ=0.8201 params={'learning_rate': 0.0004035684210100053, 'steps_per_switch': 16067, 'total_timesteps': 208387}
[2026-04-15 05:15:53] [Wave4] Proposed params: {'learning_rate': 0.0003927960467617446, 'steps_per_switch': 19892, 'total_timesteps': 201785}
[2026-04-15 05:15:55] [Wave4] Launching trial 5: {'learning_rate': 0.0003927960467617446, 'steps_per_switch': 19892, 'total_timesteps': 201785}
[2026-04-15 05:15:55] [Wave4] Command: python3 /home/paulh/projects/donkeycar-rl-autoresearch/agent/multitrack_runner.py --total-timesteps 201785 --steps-per-switch 19892 --learning-rate 0.0003927960467617446 --eval-episodes 3 --save-dir /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/wave4-trial-0005
[2026-04-15 07:15:57] [Wave4] Trial 5 TIMED OUT — killing runner.
[2026-04-15 07:15:57] [Wave4] Trial 5 finished in 7202.3s, rc=-9
[2026-04-15 07:15:57] [Wave4] Parsed: combined=None mini_monaco=None
[2026-04-15 07:15:57] [Wave4] ⚠️ No test score parsed — defaulting to 0.0
[2026-04-15 07:15:57] [Wave4] combined_test_score=0 — excluded from GP (crash/timeout).
[2026-04-15 07:15:57] [Wave4] ===== Trial 5 Summary =====
[2026-04-15 07:15:57] GP data points : 3
[2026-04-15 07:15:57] Wave4 Champion: trial=3 score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 157743}
[2026-04-15 07:15:57] Top 5:
[2026-04-15 07:15:57] score=1943.10 params={'learning_rate': 0.0006852550685205609, 'steps_per_switch': 17499, 'total_timesteps': 157743}
[2026-04-15 07:15:57] score=222.07 params={'learning_rate': 0.001, 'steps_per_switch': 6000, 'total_timesteps': 80000}
[2026-04-15 07:15:57] score=45.67 params={'learning_rate': 0.0003, 'steps_per_switch': 6000, 'total_timesteps': 80000}