Qwen3.5-27B · scored

MTP k=2

reference mtp judged submitted 10-23 11:41 UTC

Passed every gate.

SpecScore1.279geometric mean of the three speedups

solo1.64× one lane; passes 1.65, 1.64, 1.53

team1.30× team lanes busy; passes 1.26, 1.34, 1.30

busy0.98× KV cache nearly full; passes 0.90, 1.02, 0.98

Lossless gate

what each check is →

Check	Measured	Limit
device memory	162.9 GiB	≤ 165.2 GiB	pass
fixed texts' log-probabilities	0.013	≤ 0.017	pass
sampled replies' likelihood	-0.0070 ± 0.0484	within ±0.02 at 95%	pass
share of first choices	+0.0010 ± 0.0112	within ±0.005 at 95%	pass
greedy agreement	0.543	none on this task	n/a
reply lengths	–	1 ± 0.05 at 95%	pass

Speculation

Workload	Tokens / verify step	Speedup
solo	2.58	1.64
team	2.54	1.30
busy	2.55	0.98

Speculator: none
Engine: uploaded with the submission

Everything the judge recorded

{
 "ratios": {
  "solo": {
   "passes": [
    1.6478611914689827,
    1.6427288531639885,
    1.5257838056905135
   ],
   "ratio": 1.6427288531639885
  },
  "team": {
   "passes": [
    1.2644923090573588,
    1.34463957205578,
    1.2992040087185703
   ],
   "ratio": 1.2992040087185703
  },
  "busy": {
   "passes": [
    0.8973730829849887,
    1.0187307583797276,
    0.9801964431853212
   ],
   "ratio": 0.9801964431853212
  }
 },
 "specscore": 1.278945740373092,
 "greedy_agreement": 0.5433180302011483,
 "likelihood": {
  "gap": -0.007044557063254908,
  "se": 0.02471347809339398,
  "candidate": -0.48337158882580245,
  "stock": -0.47632703176254754
 },
 "argmax": {
  "gap": 0.00101609232083677,
  "se": 0.005712662165394912,
  "candidate": 0.13195177332182034,
  "stock": 0.13093568100098357
 },
 "fidelity": 0.012527069215438691,
 "lengths": {
  "solo": {
   "ratio": 1.2020018575155436,
   "low": 0.7181105372896764,
   "high": 1.6858931777414106
  },
  "team": {
   "ratio": 0.9507099131839076,
   "low": 0.7077874983443719,
   "high": 1.1936323280234433
  },
  "busy": {
   "ratio": 1.0489252639658246,
   "low": 0.7863804243298774,
   "high": 1.311470103601772
  }
 },
 "memory": {
  "ok": true,
  "peak_mib": 166840.0,
  "allowed_mib": 169119.1
 },
 "acceptance": {
  "solo": 2.5832059273197694,
  "team": 2.5354475357134225,
  "busy": 2.5526244892461047
 },
 "limits": {
  "greedy_agreement_floor": null,
  "likelihood_bound": 0.02,
  "argmax_bound": 0.005,
  "fidelity_ceiling": 0.017,
  "length_margin": 0.05,
  "memory_slack_mib": 4096
 },
 "reason": null,
 "score": 1.278945740373092
}