diff options
| author | Yuren Hao <yurenh2@illinois.edu> | 2026-07-05 05:12:34 -0500 |
|---|---|---|
| committer | Yuren Hao <yurenh2@illinois.edu> | 2026-07-05 05:12:34 -0500 |
| commit | 5975caf0f52355276aabec5ba8f144db9ac46619 (patch) | |
| tree | c4ee33f44a765ecdd31f0ce57421301bef52d211 /ep_run/r_sweep.log | |
| parent | 75ff326dcb40cd960abd56e9c9c18a45d9e5e2c2 (diff) | |
r-sweep: cos(EP,BPTT) FLAT across r=0.02..0.4 on both operators — nudge-SNR hypothesis refuted
The 50% single-shot sensitivity is MULTIPLICATIVE (near-marginal T2 dynamics'
state-sensitivity; signal and error scale with r together) — not additive
noise divided by 2r. cos ceiling ~0.88 at near-edge states is set by T2
truncation + state sensitivity (0.98 at deeply-contracted states). hr-0.2's
empirical wins are NOT estimator SNR. Hardware upside: algorithm indifferent
to r across 20x -> nudge amplitude free to fight ADDITIVE readout noise.
Surviving accuracy levers: kappa nbrake (Tikhonov, targets the real culprit),
tail-window averaging over argmin selection.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014FAPDWQ49M5Ye3NpTndTpn
Diffstat (limited to 'ep_run/r_sweep.log')
| -rw-r--r-- | ep_run/r_sweep.log | 15 |
1 files changed, 15 insertions, 0 deletions
diff --git a/ep_run/r_sweep.log b/ep_run/r_sweep.log new file mode 100644 index 0000000..057b0eb --- /dev/null +++ b/ep_run/r_sweep.log @@ -0,0 +1,15 @@ +=== s2000 === +/home/yurenh2/miniconda3/lib/python3.13/site-packages/torch/autograd/graph.py:865: UserWarning: Attempting to run cuBLAS, but there was no current CUDA context! Attempting to set the primary context... (Triggered internally at /pytorch/aten/src/ATen/cuda/CublasHandlePool.cpp:330.) + return Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass + r=0.02 cos=0.9068 0.8533 mean=0.8801 + r=0.05 cos=0.9127 0.8550 mean=0.8839 + r=0.1 cos=0.9129 0.8573 mean=0.8851 + r=0.2 cos=0.9127 0.8573 mean=0.8850 + r=0.4 cos=0.9131 0.8573 mean=0.8852 +=== fast-adaptive@2.18 === + r=0.02 cos=0.9242 0.8212 mean=0.8727 + r=0.05 cos=0.9245 0.8213 mean=0.8729 + r=0.1 cos=0.9242 0.8213 mean=0.8728 + r=0.2 cos=0.9242 0.8213 mean=0.8728 + r=0.4 cos=0.9243 0.8212 mean=0.8727 +R_SWEEP_DONE |