shanto268 commited on
Commit
56068de
·
verified ·
1 Parent(s): c167066

sync from registry v1

Browse files
Files changed (2) hide show
  1. LIMITS.md +24 -3
  2. registry.yaml +3 -3
LIMITS.md CHANGED
@@ -87,9 +87,23 @@ families): use `self_from_rows=True` when a strictly consistent Maxwell matrix i
87
 
88
  ## 4. Deployed Spaces (HF free tier, cpu-basic)
89
 
90
- Not measured yet: the push to the Hub was not performed (blocked pending the user's explicit approval). After
91
- `python deploy/sync.py --push`, run `python deploy/tests/test_app.py --url SQuADDS/PlanarCapacitanceModel` (HF_TOKEN in
92
- the environment for private Spaces) and the same stress cases through gradio_client, and fill this section.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
93
 
94
  ## 5. Known biases and reporting (review fixes, 2026-10-08)
95
 
@@ -108,6 +122,13 @@ the environment for private Spaces) and the same stress cases through gradio_cli
108
 
109
  ## 6. Flagship tiers: `all-data-v3s` (coarse prior, default) and `all-data-v3-rich` (fine prior), 2-thread latency
110
 
 
 
 
 
 
 
 
111
  Naming (2026-10-09): **coarse prior** = one gap/3 solve on the conductor-anchored v3 mesh (was "fast"), **fine prior** =
112
  gap/3 + gap/6 with Richardson extrapolation (was "accurate"); `predict(model="coarse" | "fine")`, the old words are kept as
113
  aliases and in the model ids. The chip-anchored baselines are the earlier standard (8k) and earlier 16k priors (the latter
 
87
 
88
  ## 4. Deployed Spaces (HF free tier, cpu-basic)
89
 
90
+ Measured 2026-10-10 after the fine-tier switch to `all-data-flagship` (private Spaces, `examples/client.py` path
91
+ `/predict_gds`, one request at a time; server time = prepare + predict inside the Space; wall adds upload / queue). Every
92
+ returned matrix equals the local prediction of the same bundle to <= 3e-5 relative (max over entries).
93
+
94
+ | layout | coarse tier (`all-data-v3s`), server s | fine tier (`all-data-flagship`), server s | note |
95
+ |---|---|---|---|
96
+ | Xmon example | 1.2 | 1.5 | |
97
+ | interdigital example | 2.4 | 0.8 | |
98
+ | TransmonPocket example | 19.7 (routed to the fine tier) | 19.4 | capped coarse mesh < 0.38 panels/gap: large-layout routing |
99
+ | shielded concentric example | 4.8 | 79.6 | flagship H-matrix solve, Mac 1 thread 14.6 s |
100
+ | two pads, size 20x (stress) | 61.7 (routed) | 60.2 | |
101
+ | comb, 200 fingers (stress) | 224.4 (routed) | 220.0 | |
102
+
103
+ The Space CPU is 3.4-5.4x slower than the Mac at 1 thread on the flagship solves (comb 200 fingers 64.7 s, two pads 20x 13.5 s, shielded concentric 14.6 s locally). Paper-models Space: same matrices and times
104
+ (e.g. TransmonPocket 18.0 s fine). Gradio has no request timeout; the queue runs 2 requests at once (`TLSQ_CONCURRENCY`)
105
+ but at most one flagship-prior solve (`TLSQ_HEAVY_CONCURRENCY`, default 1; fine tier and coarse requests routed to it),
106
+ so two large solves (up to 8.3 GB each, grid_qubit_9) cannot exceed the 16 GB of cpu-basic.
107
 
108
  ## 5. Known biases and reporting (review fixes, 2026-10-08)
109
 
 
122
 
123
  ## 6. Flagship tiers: `all-data-v3s` (coarse prior, default) and `all-data-v3-rich` (fine prior), 2-thread latency
124
 
125
+ **Update 2026-10-10:** the fine tier is now `all-data-flagship` (same network as `all-data-v3-rich`, flagship prior;
126
+ deploy/MODEL_SELECTION.md Results 3, PRIOR_SCHEME.md section 8.1). Mac M5 Pro, 2 threads, whole `predict`: median /
127
+ p90 / max 0.87 / 11.6 / 14.0 s on the 212 held-out designs and 1.51 / 27.6 / 78.4 s on 1,158 held-out designs (former
128
+ fine tier 4.64 / 8.96 / 18.0 s); peak RSS <= 8.3 GB on the stress and grid layouts (grid_qubit_9, 220k panels), every
129
+ one returns a matrix (the resolution floor L/8192 limits comb_f400 / pads_s20 / pads_s50 to 4.65 / 4.64 / 1.97 panels
130
+ per gap). The table below is the former fine tier (`all-data-v3-rich`) and stays as its record.
131
+
132
  Naming (2026-10-09): **coarse prior** = one gap/3 solve on the conductor-anchored v3 mesh (was "fast"), **fine prior** =
133
  gap/3 + gap/6 with Richardson extrapolation (was "accurate"); `predict(model="coarse" | "fine")`, the old words are kept as
134
  aliases and in the model ids. The chip-anchored baselines are the earlier standard (8k) and earlier 16k priors (the latter
registry.yaml CHANGED
@@ -94,9 +94,9 @@ defaults:
94
  tiers: # shown in the flagship UI; keys = tier names (aliases fast -> coarse, accurate -> fine still accepted by predict);
95
  # held-out = OOD-60 + pocket3 random-36 + Qiskit Metal new 116, pooled mutual entries
96
  coarse: {latency_s: 4, median_pct: 1.75, mae_ff: 0.534, rmse_ff: 1.173, set: "212 held-out designs (OOD-60, random TransmonPocket-3, 6 new Qiskit Metal classes)",
97
- note: "On this free CPU Space: 2.5-6 s for single components (measured on the 4 examples, refactored code); very long interdigitated combs (hundreds of fingers) can take several minutes."}
98
- fine: {latency_s: 5, median_pct: 1.13, mae_ff: 0.374, rmse_ff: 0.836, set: "212 held-out designs (OOD-60, random TransmonPocket-3, 6 new Qiskit Metal classes)",
99
- note: "Flagship prior: strip-merged Richardson (6 and 3 panels across the smallest gap on every layout), H-matrix solve for large systems. Mac, 2 threads: median 0.9 s, p90 11.6 s on the 212 held-out designs (p90 27.6 s over 1,158 incl. large shielded concentric components); this free CPU Space is slower. One fine-tier request runs at a time."}
100
 
101
  models:
102
  # ---------------------------------------------------------------- learned, released
 
94
  tiers: # shown in the flagship UI; keys = tier names (aliases fast -> coarse, accurate -> fine still accepted by predict);
95
  # held-out = OOD-60 + pocket3 random-36 + Qiskit Metal new 116, pooled mutual entries
96
  coarse: {latency_s: 4, median_pct: 1.75, mae_ff: 0.534, rmse_ff: 1.173, set: "212 held-out designs (OOD-60, random TransmonPocket-3, 6 new Qiskit Metal classes)",
97
+ note: "On this free CPU Space: 1-5 s for single components (measured on the examples, 2026-10-10). Layouts whose coarse mesh would fall below the training resolution (e.g. the TransmonPocket example, long combs) are predicted with the fine tier instead (reported under model used)."}
98
+ fine: {latency_s: 10, median_pct: 1.13, mae_ff: 0.374, rmse_ff: 0.836, set: "212 held-out designs (OOD-60, random TransmonPocket-3, 6 new Qiskit Metal classes)",
99
+ note: "Flagship prior: strip-merged Richardson (6 and 3 panels across the smallest gap on every layout), H-matrix solve for large systems. On this free CPU Space (2026-10-10): 1-2 s for the Xmon and interdigital examples, 20 s for the TransmonPocket, 80 s for the shielded concentric example; large layouts several minutes (200-finger comb 220 s). One fine-tier solve runs at a time; other requests are not blocked."}
100
 
101
  models:
102
  # ---------------------------------------------------------------- learned, released