Эксперимент: Fast-Class MoE Router (#1249)

{ "worked": false, "confidence": 9, "verdict": "Built a self-contained Fast-Class MoE router prototype with rate-scaled generators, stationary estimates, effective class-rate reduction, and a synthetic top-1 routing comparison. The Markov-chain tree theorem matched the stationary solve to 6.6e-15, and fast-class reduction error decreased from 2.09e-3 at gamma=10 to 2.56e-7 at gamma=1e5. However, the idea only marginally reduced churn while worsening overflow and load CV, so the promised practical win was not observed.", "metrics": { "baseline": "churn=0.857915, overflow=0.335078, load_CV=0.943450, domain-group accuracy=0.824336", "idea": "churn=0.856297, overflow=0.347148, load_CV=0.963904, domain-group accuracy=0.842461; stationary reduction error=2.56e-7 at gamma=1e5" }, "how_to_run": "/home/maxwelhelp/main/bin/python3 experiment.py", "files": [ "experiment.py", "results.json" ], "limitations": "Only a small synthetic top-1 router was tested; no learned pairwise router, top-2 capacity-constrained dispatch, backpropagation, transformer/C4 training, equal-FLOP perplexity comparison, communication measurement, or differentiable stationary estimation was evaluated." }