diff --git a/agent/models/champion/manifest.json b/agent/models/champion/manifest.json new file mode 100644 index 0000000..1b9bfb4 --- /dev/null +++ b/agent/models/champion/manifest.json @@ -0,0 +1,15 @@ +{ + "trial": 8, + "timestamp": "2026-04-13T11:52:48.821545", + "params": { + "n_steer": 6, + "n_throttle": 2, + "learning_rate": 0.001449588903551847, + "timesteps": 3429, + "agent": "ppo", + "eval_episodes": 3, + "reward_shaping": true + }, + "mean_reward": 1936.8533, + "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/champion/model.zip" +} \ No newline at end of file diff --git a/agent/outerloop-results/autoresearch_phase1_log.txt b/agent/outerloop-results/autoresearch_phase1_log.txt index 2ab8a04..c103eb2 100644 --- a/agent/outerloop-results/autoresearch_phase1_log.txt +++ b/agent/outerloop-results/autoresearch_phase1_log.txt @@ -105,3 +105,199 @@ [2026-04-13 11:16:53] [Champion] 🏆 NEW BEST! Trial 3: mean_reward=90.0000 params={'r': 90} [2026-04-13 11:16:53] [Champion] 🏆 NEW BEST! Trial 5: mean_reward=75.0000 params={'n_steer': 8} [2026-04-13 11:16:53] [AutoResearch] Only 1 results — using random proposal. +[2026-04-13 11:17:15] ============================================================ +[2026-04-13 11:17:15] [AutoResearch] Phase 1 — Real PPO Training + GP+UCB Optimization +[2026-04-13 11:17:15] [AutoResearch] Max trials: 50 | kappa: 2.0 | push every: 10 +[2026-04-13 11:17:15] [AutoResearch] Results: /home/paulh/projects/donkeycar-rl-autoresearch/agent/outerloop-results/autoresearch_results_phase1.jsonl +[2026-04-13 11:17:15] [AutoResearch] Champion: /home/paulh/projects/donkeycar-rl-autoresearch/agent/models/champion +[2026-04-13 11:17:15] ============================================================ +[2026-04-13 11:17:15] [AutoResearch] Loaded 0 existing Phase 1 results. +[2026-04-13 11:17:15] [AutoResearch] No champion yet. +[2026-04-13 11:17:15] +[AutoResearch] ========== Trial 1/50 ========== +[2026-04-13 11:17:15] [AutoResearch] Only 0 results — using random proposal. +[2026-04-13 11:17:15] [AutoResearch] Proposed: {'n_steer': 4, 'n_throttle': 4, 'learning_rate': 0.0033894181299375602, 'timesteps': 2116, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:17:17] [AutoResearch] Launching trial 1: {'n_steer': 4, 'n_throttle': 4, 'learning_rate': 0.0033894181299375602, 'timesteps': 2116, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:20:53] [AutoResearch] Trial 1 finished in 215.5s, returncode=0 +[2026-04-13 11:20:53] [AutoResearch] Trial 1: mean_reward=5.7246 std_reward=0.027 +[2026-04-13 11:20:53] [Champion] 🏆 NEW BEST! Trial 1: mean_reward=5.7246 params={'n_steer': 4, 'n_throttle': 4, 'learning_rate': 0.0033894181299375602, 'timesteps': 2116, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:20:53] [AutoResearch] === Trial 1 Summary === +[2026-04-13 11:20:53] Total Phase 1 runs: 1 +[2026-04-13 11:20:53] Champion: trial=1 mean_reward=5.7246 params={'n_steer': 4, 'n_throttle': 4, 'learning_rate': 0.0033894181299375602, 'timesteps': 2116, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:20:53] Top 5: +[2026-04-13 11:20:53] mean_reward=5.7246 params={'n_steer': 4, 'n_throttle': 4, 'learning_rate': 0.0033894181299375602, 'timesteps': 2116, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:20:55] +[AutoResearch] ========== Trial 2/50 ========== +[2026-04-13 11:20:55] [AutoResearch] Only 1 results — using random proposal. +[2026-04-13 11:20:55] [AutoResearch] Proposed: {'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:20:57] [AutoResearch] Launching trial 2: {'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:24:32] [AutoResearch] Trial 2 finished in 215.5s, returncode=0 +[2026-04-13 11:24:32] [AutoResearch] Trial 2: mean_reward=398.8564 std_reward=1.1786 +[2026-04-13 11:24:33] [Champion] 🏆 NEW BEST! Trial 2: mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:24:33] [AutoResearch] === Trial 2 Summary === +[2026-04-13 11:24:33] Total Phase 1 runs: 2 +[2026-04-13 11:24:33] Champion: trial=2 mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:24:33] Top 5: +[2026-04-13 11:24:33] mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:24:33] mean_reward=5.7246 params={'n_steer': 4, 'n_throttle': 4, 'learning_rate': 0.0033894181299375602, 'timesteps': 2116, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:24:35] +[AutoResearch] ========== Trial 3/50 ========== +[2026-04-13 11:24:35] [AutoResearch] Only 2 results — using random proposal. +[2026-04-13 11:24:35] [AutoResearch] Proposed: {'n_steer': 7, 'n_throttle': 4, 'learning_rate': 0.003869851995924804, 'timesteps': 2985, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:24:37] [AutoResearch] Launching trial 3: {'n_steer': 7, 'n_throttle': 4, 'learning_rate': 0.003869851995924804, 'timesteps': 2985, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:29:06] [AutoResearch] Trial 3 finished in 269.6s, returncode=0 +[2026-04-13 11:29:06] [AutoResearch] Trial 3: mean_reward=5.9776 std_reward=0.0252 +[2026-04-13 11:29:06] [AutoResearch] === Trial 3 Summary === +[2026-04-13 11:29:06] Total Phase 1 runs: 3 +[2026-04-13 11:29:06] Champion: trial=2 mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:29:06] Top 5: +[2026-04-13 11:29:06] mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:29:06] mean_reward=5.9776 params={'n_steer': 7, 'n_throttle': 4, 'learning_rate': 0.003869851995924804, 'timesteps': 2985, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:29:06] mean_reward=5.7246 params={'n_steer': 4, 'n_throttle': 4, 'learning_rate': 0.0033894181299375602, 'timesteps': 2116, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:29:08] +[AutoResearch] ========== Trial 4/50 ========== +[2026-04-13 11:29:08] [AutoResearch] GP UCB top-5 candidates: +[2026-04-13 11:29:08] UCB=2.4615 mu=0.8615 sigma=0.8000 params={'n_steer': 8, 'n_throttle': 2, 'learning_rate': 0.0010468048869752956, 'timesteps': 3084} +[2026-04-13 11:29:08] UCB=2.4548 mu=0.9032 sigma=0.7758 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.0009758380297094257, 'timesteps': 3202} +[2026-04-13 11:29:08] UCB=2.4540 mu=0.7444 sigma=0.8548 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.0006970147905651335, 'timesteps': 3351} +[2026-04-13 11:29:08] UCB=2.4479 mu=0.7051 sigma=0.8714 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.0009997388594600006, 'timesteps': 4139} +[2026-04-13 11:29:08] UCB=2.4443 mu=0.9374 sigma=0.7535 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.001158537723428793, 'timesteps': 3743} +[2026-04-13 11:29:08] [AutoResearch] Proposed: {'n_steer': 8, 'n_throttle': 2, 'learning_rate': 0.0010468048869752956, 'timesteps': 3084, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:29:10] [AutoResearch] Launching trial 4: {'n_steer': 8, 'n_throttle': 2, 'learning_rate': 0.0010468048869752956, 'timesteps': 3084, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:32:24] [AutoResearch] Trial 4 finished in 194.0s, returncode=0 +[2026-04-13 11:32:24] [AutoResearch] Trial 4: mean_reward=22.8241 std_reward=0.1918 +[2026-04-13 11:32:24] [AutoResearch] === Trial 4 Summary === +[2026-04-13 11:32:24] Total Phase 1 runs: 4 +[2026-04-13 11:32:24] Champion: trial=2 mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:32:24] Top 5: +[2026-04-13 11:32:24] mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:32:24] mean_reward=22.8241 params={'n_steer': 8, 'n_throttle': 2, 'learning_rate': 0.0010468048869752956, 'timesteps': 3084, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:32:24] mean_reward=5.9776 params={'n_steer': 7, 'n_throttle': 4, 'learning_rate': 0.003869851995924804, 'timesteps': 2985, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:32:24] mean_reward=5.7246 params={'n_steer': 4, 'n_throttle': 4, 'learning_rate': 0.0033894181299375602, 'timesteps': 2116, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:32:26] +[AutoResearch] ========== Trial 5/50 ========== +[2026-04-13 11:32:26] [AutoResearch] GP UCB top-5 candidates: +[2026-04-13 11:32:26] UCB=2.9797 mu=1.4209 sigma=0.7794 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.003486801052946445, 'timesteps': 3626} +[2026-04-13 11:32:26] UCB=2.9360 mu=1.6516 sigma=0.6422 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.003483580964392729, 'timesteps': 3613} +[2026-04-13 11:32:26] UCB=2.8856 mu=1.1888 sigma=0.8484 params={'n_steer': 5, 'n_throttle': 2, 'learning_rate': 0.002515064142571671, 'timesteps': 4267} +[2026-04-13 11:32:26] UCB=2.8582 mu=1.5163 sigma=0.6709 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.0029159438252188284, 'timesteps': 3730} +[2026-04-13 11:32:26] UCB=2.8422 mu=1.5296 sigma=0.6563 params={'n_steer': 8, 'n_throttle': 2, 'learning_rate': 0.0033924318546103937, 'timesteps': 3346} +[2026-04-13 11:32:26] [AutoResearch] Proposed: {'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.003486801052946445, 'timesteps': 3626, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:32:28] [AutoResearch] Launching trial 5: {'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.003486801052946445, 'timesteps': 3626, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:36:52] [AutoResearch] Trial 5 finished in 264.3s, returncode=0 +[2026-04-13 11:36:52] [AutoResearch] Trial 5: mean_reward=5.9913 std_reward=0.0246 +[2026-04-13 11:36:52] [AutoResearch] === Trial 5 Summary === +[2026-04-13 11:36:52] Total Phase 1 runs: 5 +[2026-04-13 11:36:52] Champion: trial=2 mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:36:52] Top 5: +[2026-04-13 11:36:52] mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:36:52] mean_reward=22.8241 params={'n_steer': 8, 'n_throttle': 2, 'learning_rate': 0.0010468048869752956, 'timesteps': 3084, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:36:52] mean_reward=5.9913 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.003486801052946445, 'timesteps': 3626, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:36:52] mean_reward=5.9776 params={'n_steer': 7, 'n_throttle': 4, 'learning_rate': 0.003869851995924804, 'timesteps': 2985, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:36:52] mean_reward=5.7246 params={'n_steer': 4, 'n_throttle': 4, 'learning_rate': 0.0033894181299375602, 'timesteps': 2116, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:36:54] +[AutoResearch] ========== Trial 6/50 ========== +[2026-04-13 11:36:54] [AutoResearch] GP UCB top-5 candidates: +[2026-04-13 11:36:54] UCB=2.8622 mu=1.4083 sigma=0.7270 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.0023577235727246376, 'timesteps': 4387} +[2026-04-13 11:36:54] UCB=2.7841 mu=1.0518 sigma=0.8661 params={'n_steer': 8, 'n_throttle': 2, 'learning_rate': 0.002782960062629981, 'timesteps': 4433} +[2026-04-13 11:36:54] UCB=2.7380 mu=1.5849 sigma=0.5765 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.001906612836342622, 'timesteps': 3714} +[2026-04-13 11:36:54] UCB=2.7029 mu=0.9236 sigma=0.8897 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002301914433902658, 'timesteps': 4751} +[2026-04-13 11:36:54] UCB=2.6924 mu=1.1628 sigma=0.7648 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.0019575950790335435, 'timesteps': 2548} +[2026-04-13 11:36:54] [AutoResearch] Proposed: {'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.0023577235727246376, 'timesteps': 4387, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:36:56] [AutoResearch] Launching trial 6: {'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.0023577235727246376, 'timesteps': 4387, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:44:57] [AutoResearch] Trial 6 TIMED OUT after 480.1s +[2026-04-13 11:44:57] [AutoResearch] Trial 6: mean_reward=None std_reward=None +[2026-04-13 11:44:57] [AutoResearch] === Trial 6 Summary === +[2026-04-13 11:44:57] Total Phase 1 runs: 5 +[2026-04-13 11:44:57] Champion: trial=2 mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:44:57] Top 5: +[2026-04-13 11:44:57] mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:44:57] mean_reward=22.8241 params={'n_steer': 8, 'n_throttle': 2, 'learning_rate': 0.0010468048869752956, 'timesteps': 3084, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:44:57] mean_reward=5.9913 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.003486801052946445, 'timesteps': 3626, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:44:57] mean_reward=5.9776 params={'n_steer': 7, 'n_throttle': 4, 'learning_rate': 0.003869851995924804, 'timesteps': 2985, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:44:57] mean_reward=5.7246 params={'n_steer': 4, 'n_throttle': 4, 'learning_rate': 0.0033894181299375602, 'timesteps': 2116, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:44:59] +[AutoResearch] ========== Trial 7/50 ========== +[2026-04-13 11:44:59] [AutoResearch] GP UCB top-5 candidates: +[2026-04-13 11:44:59] UCB=2.7677 mu=1.3945 sigma=0.6866 params={'n_steer': 8, 'n_throttle': 3, 'learning_rate': 0.002636854645150246, 'timesteps': 3160} +[2026-04-13 11:44:59] UCB=2.6401 mu=0.8590 sigma=0.8906 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.0012329470317109907, 'timesteps': 4404} +[2026-04-13 11:44:59] UCB=2.6346 mu=0.8897 sigma=0.8725 params={'n_steer': 8, 'n_throttle': 2, 'learning_rate': 0.002824572687677801, 'timesteps': 2173} +[2026-04-13 11:44:59] UCB=2.6197 mu=1.1406 sigma=0.7395 params={'n_steer': 7, 'n_throttle': 3, 'learning_rate': 0.002264095441698803, 'timesteps': 3610} +[2026-04-13 11:44:59] UCB=2.6013 mu=0.7257 sigma=0.9378 params={'n_steer': 8, 'n_throttle': 3, 'learning_rate': 0.001986372556075669, 'timesteps': 4899} +[2026-04-13 11:44:59] [AutoResearch] Proposed: {'n_steer': 8, 'n_throttle': 3, 'learning_rate': 0.002636854645150246, 'timesteps': 3160, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:45:01] [AutoResearch] Launching trial 7: {'n_steer': 8, 'n_throttle': 3, 'learning_rate': 0.002636854645150246, 'timesteps': 3160, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:48:32] [AutoResearch] Trial 7 finished in 211.3s, returncode=0 +[2026-04-13 11:48:32] [AutoResearch] Trial 7: mean_reward=5.7529 std_reward=0.0318 +[2026-04-13 11:48:32] [AutoResearch] === Trial 7 Summary === +[2026-04-13 11:48:32] Total Phase 1 runs: 6 +[2026-04-13 11:48:32] Champion: trial=2 mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:48:32] Top 5: +[2026-04-13 11:48:32] mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:48:32] mean_reward=22.8241 params={'n_steer': 8, 'n_throttle': 2, 'learning_rate': 0.0010468048869752956, 'timesteps': 3084, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:48:32] mean_reward=5.9913 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.003486801052946445, 'timesteps': 3626, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:48:32] mean_reward=5.9776 params={'n_steer': 7, 'n_throttle': 4, 'learning_rate': 0.003869851995924804, 'timesteps': 2985, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:48:32] mean_reward=5.7529 params={'n_steer': 8, 'n_throttle': 3, 'learning_rate': 0.002636854645150246, 'timesteps': 3160, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:48:34] +[AutoResearch] ========== Trial 8/50 ========== +[2026-04-13 11:48:34] [AutoResearch] GP UCB top-5 candidates: +[2026-04-13 11:48:34] UCB=2.9928 mu=1.4031 sigma=0.7948 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.001449588903551847, 'timesteps': 3429} +[2026-04-13 11:48:34] UCB=2.9102 mu=1.2105 sigma=0.8499 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.0013337121696681005, 'timesteps': 4384} +[2026-04-13 11:48:34] UCB=2.9095 mu=1.2362 sigma=0.8366 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.0016866172466465327, 'timesteps': 4586} +[2026-04-13 11:48:34] UCB=2.7220 mu=1.0017 sigma=0.8601 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.0012033218829906316, 'timesteps': 4239} +[2026-04-13 11:48:34] UCB=2.6586 mu=0.8020 sigma=0.9283 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.0014425382569831862, 'timesteps': 4488} +[2026-04-13 11:48:34] [AutoResearch] Proposed: {'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.001449588903551847, 'timesteps': 3429, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:48:36] [AutoResearch] Launching trial 8: {'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.001449588903551847, 'timesteps': 3429, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:52:48] [AutoResearch] Trial 8 finished in 252.2s, returncode=0 +[2026-04-13 11:52:48] [AutoResearch] Trial 8: mean_reward=1936.8533 std_reward=34.0067 +[2026-04-13 11:52:48] [Champion] 🏆 NEW BEST! Trial 8: mean_reward=1936.8533 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.001449588903551847, 'timesteps': 3429, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:52:48] [AutoResearch] === Trial 8 Summary === +[2026-04-13 11:52:48] Total Phase 1 runs: 7 +[2026-04-13 11:52:48] Champion: trial=8 mean_reward=1936.8533 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.001449588903551847, 'timesteps': 3429, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:52:48] Top 5: +[2026-04-13 11:52:48] mean_reward=1936.8533 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.001449588903551847, 'timesteps': 3429, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:52:48] mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:52:48] mean_reward=22.8241 params={'n_steer': 8, 'n_throttle': 2, 'learning_rate': 0.0010468048869752956, 'timesteps': 3084, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:52:48] mean_reward=5.9913 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.003486801052946445, 'timesteps': 3626, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:52:48] mean_reward=5.9776 params={'n_steer': 7, 'n_throttle': 4, 'learning_rate': 0.003869851995924804, 'timesteps': 2985, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:52:50] +[AutoResearch] ========== Trial 9/50 ========== +[2026-04-13 11:52:50] [AutoResearch] GP UCB top-5 candidates: +[2026-04-13 11:52:50] UCB=3.6446 mu=2.2362 sigma=0.7042 params={'n_steer': 4, 'n_throttle': 2, 'learning_rate': 0.0012562469886511318, 'timesteps': 2961} +[2026-04-13 11:52:50] UCB=3.6253 mu=2.3605 sigma=0.6324 params={'n_steer': 5, 'n_throttle': 2, 'learning_rate': 0.0014035502090066865, 'timesteps': 2985} +[2026-04-13 11:52:50] UCB=3.5079 mu=2.3661 sigma=0.5709 params={'n_steer': 5, 'n_throttle': 2, 'learning_rate': 0.0016891347290568105, 'timesteps': 3422} +[2026-04-13 11:52:50] UCB=3.4169 mu=2.2243 sigma=0.5963 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.0011351472472032882, 'timesteps': 4191} +[2026-04-13 11:52:50] UCB=3.3399 mu=1.6131 sigma=0.8634 params={'n_steer': 4, 'n_throttle': 2, 'learning_rate': 0.00114114991914373, 'timesteps': 3413} +[2026-04-13 11:52:50] [AutoResearch] Proposed: {'n_steer': 4, 'n_throttle': 2, 'learning_rate': 0.0012562469886511318, 'timesteps': 2961, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:52:52] [AutoResearch] Launching trial 9: {'n_steer': 4, 'n_throttle': 2, 'learning_rate': 0.0012562469886511318, 'timesteps': 2961, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:56:28] [AutoResearch] Trial 9 finished in 215.5s, returncode=0 +[2026-04-13 11:56:28] [AutoResearch] Trial 9: mean_reward=237.9115 std_reward=1.4136 +[2026-04-13 11:56:28] [AutoResearch] === Trial 9 Summary === +[2026-04-13 11:56:28] Total Phase 1 runs: 8 +[2026-04-13 11:56:28] Champion: trial=8 mean_reward=1936.8533 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.001449588903551847, 'timesteps': 3429, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:56:28] Top 5: +[2026-04-13 11:56:28] mean_reward=1936.8533 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.001449588903551847, 'timesteps': 3429, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:56:28] mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:56:28] mean_reward=237.9115 params={'n_steer': 4, 'n_throttle': 2, 'learning_rate': 0.0012562469886511318, 'timesteps': 2961, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:56:28] mean_reward=22.8241 params={'n_steer': 8, 'n_throttle': 2, 'learning_rate': 0.0010468048869752956, 'timesteps': 3084, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:56:28] mean_reward=5.9913 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.003486801052946445, 'timesteps': 3626, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:56:30] +[AutoResearch] ========== Trial 10/50 ========== +[2026-04-13 11:56:30] [AutoResearch] GP UCB top-5 candidates: +[2026-04-13 11:56:30] UCB=3.6513 mu=2.0026 sigma=0.8243 params={'n_steer': 5, 'n_throttle': 2, 'learning_rate': 0.0012074041487018196, 'timesteps': 4691} +[2026-04-13 11:56:30] UCB=3.2438 mu=1.9644 sigma=0.6397 params={'n_steer': 5, 'n_throttle': 2, 'learning_rate': 0.0013292739097783752, 'timesteps': 3897} +[2026-04-13 11:56:30] UCB=3.1815 mu=1.2984 sigma=0.9415 params={'n_steer': 6, 'n_throttle': 3, 'learning_rate': 0.0004768119261475519, 'timesteps': 4914} +[2026-04-13 11:56:30] UCB=3.0779 mu=1.4273 sigma=0.8253 params={'n_steer': 6, 'n_throttle': 3, 'learning_rate': 0.0001854044179957165, 'timesteps': 3308} +[2026-04-13 11:56:30] UCB=2.9649 mu=1.2760 sigma=0.8444 params={'n_steer': 6, 'n_throttle': 3, 'learning_rate': 0.0001236030774697938, 'timesteps': 3010} +[2026-04-13 11:56:30] [AutoResearch] Proposed: {'n_steer': 5, 'n_throttle': 2, 'learning_rate': 0.0012074041487018196, 'timesteps': 4691, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 11:56:32] [AutoResearch] Launching trial 10: {'n_steer': 5, 'n_throttle': 2, 'learning_rate': 0.0012074041487018196, 'timesteps': 4691, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 12:01:17] [AutoResearch] Trial 10 finished in 285.4s, returncode=0 +[2026-04-13 12:01:17] [AutoResearch] Trial 10: mean_reward=7.6595 std_reward=0.1051 +[2026-04-13 12:01:17] [AutoResearch] === Trial 10 Summary === +[2026-04-13 12:01:17] Total Phase 1 runs: 9 +[2026-04-13 12:01:17] Champion: trial=8 mean_reward=1936.8533 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.001449588903551847, 'timesteps': 3429, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 12:01:17] Top 5: +[2026-04-13 12:01:17] mean_reward=1936.8533 params={'n_steer': 6, 'n_throttle': 2, 'learning_rate': 0.001449588903551847, 'timesteps': 3429, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 12:01:17] mean_reward=398.8564 params={'n_steer': 7, 'n_throttle': 2, 'learning_rate': 0.002359934949698355, 'timesteps': 3386, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 12:01:17] mean_reward=237.9115 params={'n_steer': 4, 'n_throttle': 2, 'learning_rate': 0.0012562469886511318, 'timesteps': 2961, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 12:01:17] mean_reward=22.8241 params={'n_steer': 8, 'n_throttle': 2, 'learning_rate': 0.0010468048869752956, 'timesteps': 3084, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} +[2026-04-13 12:01:17] mean_reward=7.6595 params={'n_steer': 5, 'n_throttle': 2, 'learning_rate': 0.0012074041487018196, 'timesteps': 4691, 'agent': 'ppo', 'eval_episodes': 3, 'reward_shaping': True} diff --git a/agent/outerloop-results/autoresearch_results_phase1.jsonl b/agent/outerloop-results/autoresearch_results_phase1.jsonl index 6ab5a20..56a5326 100644 --- a/agent/outerloop-results/autoresearch_results_phase1.jsonl +++ b/agent/outerloop-results/autoresearch_results_phase1.jsonl @@ -4,3 +4,13 @@ {"trial": 4, "timestamp": "2026-04-13T10:43:36.999174", "params": {"n_steer": 8, "n_throttle": 2, "learning_rate": 0.0005174658025335539, "timesteps": 22022, "agent": "ppo", "eval_episodes": 5, "reward_shaping": true}, "mean_reward": null, "std_reward": null, "model_path": null, "champion": false, "run_status": "timeout", "elapsed_sec": 600.093513250351} {"trial": 5, "timestamp": "2026-04-13T10:53:41.112283", "params": {"n_steer": 4, "n_throttle": 3, "learning_rate": 0.004765524064388173, "timesteps": 23582, "agent": "ppo", "eval_episodes": 5, "reward_shaping": true}, "mean_reward": null, "std_reward": null, "model_path": null, "champion": false, "run_status": "timeout", "elapsed_sec": 600.0998013019562} {"trial": 6, "timestamp": "2026-04-13T11:03:45.201524", "params": {"n_steer": 8, "n_throttle": 2, "learning_rate": 0.0008238758073115486, "timesteps": 23327, "agent": "ppo", "eval_episodes": 5, "reward_shaping": true}, "mean_reward": null, "std_reward": null, "model_path": null, "champion": false, "run_status": "timeout", "elapsed_sec": 600.0790619850159} +{"trial": 1, "timestamp": "2026-04-13T11:20:53.376401", "params": {"n_steer": 4, "n_throttle": 4, "learning_rate": 0.0033894181299375602, "timesteps": 2116, "agent": "ppo", "eval_episodes": 3, "reward_shaping": true}, "mean_reward": 5.7246, "std_reward": 0.027, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/trial-0001/model.zip", "champion": true, "run_status": "ok", "elapsed_sec": 215.45383524894714} +{"trial": 2, "timestamp": "2026-04-13T11:24:33.013533", "params": {"n_steer": 7, "n_throttle": 2, "learning_rate": 0.002359934949698355, "timesteps": 3386, "agent": "ppo", "eval_episodes": 3, "reward_shaping": true}, "mean_reward": 398.8564, "std_reward": 1.1786, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/trial-0002/model.zip", "champion": true, "run_status": "ok", "elapsed_sec": 215.54161620140076} +{"trial": 3, "timestamp": "2026-04-13T11:29:06.639561", "params": {"n_steer": 7, "n_throttle": 4, "learning_rate": 0.003869851995924804, "timesteps": 2985, "agent": "ppo", "eval_episodes": 3, "reward_shaping": true}, "mean_reward": 5.9776, "std_reward": 0.0252, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/trial-0003/model.zip", "champion": false, "run_status": "ok", "elapsed_sec": 269.6129357814789} +{"trial": 4, "timestamp": "2026-04-13T11:32:24.687658", "params": {"n_steer": 8, "n_throttle": 2, "learning_rate": 0.0010468048869752956, "timesteps": 3084, "agent": "ppo", "eval_episodes": 3, "reward_shaping": true}, "mean_reward": 22.8241, "std_reward": 0.1918, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/trial-0004/model.zip", "champion": false, "run_status": "ok", "elapsed_sec": 194.03403115272522} +{"trial": 5, "timestamp": "2026-04-13T11:36:52.972708", "params": {"n_steer": 6, "n_throttle": 2, "learning_rate": 0.003486801052946445, "timesteps": 3626, "agent": "ppo", "eval_episodes": 3, "reward_shaping": true}, "mean_reward": 5.9913, "std_reward": 0.0246, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/trial-0005/model.zip", "champion": false, "run_status": "ok", "elapsed_sec": 264.26293778419495} +{"trial": 6, "timestamp": "2026-04-13T11:44:57.105479", "params": {"n_steer": 7, "n_throttle": 2, "learning_rate": 0.0023577235727246376, "timesteps": 4387, "agent": "ppo", "eval_episodes": 3, "reward_shaping": true}, "mean_reward": null, "std_reward": null, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/trial-0006/model.zip", "champion": false, "run_status": "timeout", "elapsed_sec": 480.1191928386688} +{"trial": 7, "timestamp": "2026-04-13T11:48:32.430576", "params": {"n_steer": 8, "n_throttle": 3, "learning_rate": 0.002636854645150246, "timesteps": 3160, "agent": "ppo", "eval_episodes": 3, "reward_shaping": true}, "mean_reward": 5.7529, "std_reward": 0.0318, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/trial-0007/model.zip", "champion": false, "run_status": "ok", "elapsed_sec": 211.3090944290161} +{"trial": 8, "timestamp": "2026-04-13T11:52:48.821996", "params": {"n_steer": 6, "n_throttle": 2, "learning_rate": 0.001449588903551847, "timesteps": 3429, "agent": "ppo", "eval_episodes": 3, "reward_shaping": true}, "mean_reward": 1936.8533, "std_reward": 34.0067, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/trial-0008/model.zip", "champion": true, "run_status": "ok", "elapsed_sec": 252.2464599609375} +{"trial": 9, "timestamp": "2026-04-13T11:56:28.296244", "params": {"n_steer": 4, "n_throttle": 2, "learning_rate": 0.0012562469886511318, "timesteps": 2961, "agent": "ppo", "eval_episodes": 3, "reward_shaping": true}, "mean_reward": 237.9115, "std_reward": 1.4136, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/trial-0009/model.zip", "champion": false, "run_status": "ok", "elapsed_sec": 215.46081495285034} +{"trial": 10, "timestamp": "2026-04-13T12:01:17.700485", "params": {"n_steer": 5, "n_throttle": 2, "learning_rate": 0.0012074041487018196, "timesteps": 4691, "agent": "ppo", "eval_episodes": 3, "reward_shaping": true}, "mean_reward": 7.6595, "std_reward": 0.1051, "model_path": "/home/paulh/projects/donkeycar-rl-autoresearch/agent/models/trial-0010/model.zip", "champion": false, "run_status": "ok", "elapsed_sec": 285.3893370628357}