We examine supervised classical prediction of all one-qubit Z and two-qubit ZZ expectation values on noiseless, layered 4-10-qubit RX/RY/RZ-CNOT circuits. In a 63-cell stratified workload with disjoint circuit-template tests, we evaluate one transformer encoder and gate-dependency graph neural networks against literal-zero and training-derived constant predictors. On the Arm K training labels, 44.89% of scalar targets have absolute magnitude below 0.05. The held-out-template zero predictor has equal-instance mean absolute error 0.193399, and none of the evaluated transformer checkpoints or final graph-network endpoints achieves a lower score. A separate product-state positive control achieves MAE 0.007755 versus 0.514811 for zero, but does not explain the failure on the entangled workload. The difference between zero and a training-fitted mean constant under training MSE is only 0.14%, a descriptive point estimate without uncertainty analysis. The transformer's 815,233-parameter source identity is locally verified; detailed packed-input semantics have not been independently audited. These results constitute a descriptive pilot, not a causal explanation, formal equivalence test, or impossibility claim. Comparable circuits can be regenerated from the workload specification, but exact frozen datasets and model-run evidence remain private.
Figures & tables
Estimator
Train
Within-template
Held-out templates
Literal zero
0.168849
0.169009
0.170763
Train-fitted mean
0.169606
0.169906
0.171786
Train-fitted median
0.168833
0.169028
0.170789
Table 1: Pooled-scalar MAE for the Arm K panels, reported separately from the paper’s equal-instance primary metric. Constants were fitted only on training labels.
Arm
Templates/cell
Angles/template
Train instances
Updates
B
6
4
1,512
384
A
6
20
7,560
384
K
30
4
7,560
384
L
6
4
1,512
1,920
Table 2: Exploratory transformer data/exposure arms. All rows use the same d=128 , four-layer, four-head, FFN-512, W=0 architecture.
Regime
Parameters
Updates
Train MAE
Held-out MAE
K, 12 rounds
823,809
1,920
0.190717
0.193497
K, 12 rounds
823,809
7,680
0.190476
0.193707
K, 24 rounds
1,604,865
7,680
0.191281
0.194118
K, 40 rounds
2,646,273
7,680
0.191269
0.194155
Table 3: GateDAG endpoint scores, with the same held-out zero MAE of 0.193399 for all rows. These runs are not independent seeds.
Figure 1: Descriptive held-out equal-instance MAE. The zero-predictor level is dashed. GateDAG points are the stated run endpoints; the transformer point is the best evaluated checkpoint among 12 exploratory runs. The y-axis shows held-out equal-instance MAE (0.1932–0.1947); this narrow range is intentionally displayed to show small differences. No error bars or uncertainty inference are available for the single-seed GateDAG sweep.
Estimator
Train
Within-template
Held-out templates
Literal zero
0.095369
0.094946
0.097297
Train-fitted mean constant
0.095236
0.094958
0.097276
Table 4: Frozen-label baseline MSE under equal-instance weighting for Arm K. These are baseline metrics, not trained-model fixed-checkpoint MSE scores.
Beth Israel Deaconess Medical Center & Harvard Medical School Boston, MA, USA · Department of Electrophysics National Yang Ming Chiao Tung University Hsinchu, Taiwan