# Эксперимент: Congestion-aware equimarginal MoE router (#837)

{
  "worked": true,
  "confidence": 9,
  "verdict": "Built a congestion-aware equimarginal router using projected marginal-utility ascent with payoff backtracking, plus an independent softmax baseline. The mechanism manifested clearly: analytic support transitions matched observed transitions near budgets 0.4142, 1.4142, and 5.8284; the maximum KKT residual was 1.11e-16; and congestion-aware routing reduced expert-load CV and overflow in all five toy load sweeps. This is a promising toy signal, not evidence of an end-to-end MoE or language-model win.",
  "metrics": {
    "baseline": "Independent softmax: load CV 0.624, 0.624, 0.626, 0.611, 0.621 and overflow 0.867, 1.730, 3.380, 6.641, 13.576 for 4, 8, 16, 32, 64 groups.",
    "idea": "Equimarginal congestion routing: load CV 0.433, 0.435, 0.329, 0.458, 0.453 and overflow 0.524, 1.080, 1.331, 4.502, 8.755 for 4, 8, 16, 32, 64 groups; KKT residual 1.11e-16."
  },
  "how_to_run": "/home/maxwelhelp/main/bin/python3 router_experiment.py > results.json",
  "files": [
    "router_experiment.py",
    "results.json",
    "README.txt"
  ],
  "limitations": "Only a small fixed-seed toy allocation experiment was tested. There was no Transformer or MoE language-model training, perplexity comparison, GPU or dispatch-kernel throughput measurement, top-2 dispatch implementation, straight-through estimator, EMA capacity tracking, or large-scale capacity/drop-rate evaluation."
}