Эксперимент: Diversity-Weighted Leave-One-Out Policy Baseline (#872)

{ "worked": true, "confidence": 9, "verdict": "Built the diversity-weighted leave-one-out baseline and detached PyTorch SSPO loss. The mechanism checks matched predictions: p=0 recovered uniform LOO and effective count 5.0, distance ratios matched the squared-distance power law (16.0), and effective count fell from 4.0 to 2.00 as p increased from 0 to 2. The secondary categorical policy proxy did not show a win: diversity weighting increased gradient variance 3.47x versus uniform LOO, so this is mechanism-validated but not evidence of improved training.", "metrics": { "baseline": "Uniform LOO proxy gradient variance 0.0466", "idea": "Diversity-weighted proxy gradient variance 0.1616 (3.47x baseline); effective counts for p=0,0.5,1,2 were 4.000, 2.200, 2.010, 2.000" }, "how_to_run": "/home/maxwelhelp/main/bin/python3 run_experiment.py", "files": [ "diversity_baseline.py", "run_experiment.py", "verification.json" ], "limitations": "No neural TSP-50 solver, multi-step training curve, wall-clock throughput, final tour quality, or GPU experiment was tested. The toy variance comparison used a four-action categorical policy and does not establish general RL performance." }