diff --git a/dashboard.html b/dashboard.html index 1616749..e2d4fe7 100644 --- a/dashboard.html +++ b/dashboard.html @@ -331,11 +331,11 @@ - Open slope_hierarchy_scale: https://github.com/PyAutoLabs/slope_hierarchy_scale - Open ic50_workspace: https://github.com/Jammy2211/ic50_workspace - Open autolens_inference: https://github.com/PyAutoLabs/autolens_inference -- Open autofit_inference: https://github.com/PyAutoLabs/autofit_inference
Last updated 2026-10-15 16:00:25 UTC
↻ Update

- +- Open autofit_inference: https://github.com/PyAutoLabs/autofit_inference
Last updated 2026-10-15 17:38:42 UTC
↻ Update

+

-

Summary

22 Running · 48 Open
+

Summary

22 Running · 47 Open
@@ -345,7 +345,7 @@ - +
ProjectRunningOpenLast update
subhalo_validation052026-09-07
slope_hierarchy_scale002026-10-14
ic50_workspace002026-10-14
autolens_inference602026-09-24
autofit_inference012026-10-15
autofit_inference002026-10-15

Projects

8
@@ -550,18 +550,16 @@

autolens_inference — Do the PyAutoLens inference backends agree on the HST

autofit_inference — PyAutoFit search benchmark on gaussian_x3: which searches find the right answer fastest

active · ral · autofit_inference#8 · projects/autofit_inference.md · local /home/jammy/Code/PyAutoLabs/fit/autofit_inference · RAL /mnt/ral/jnightin/autofit_inference

Now

-

Scored wave 2 pulled and judged under gaussian_x3@2 (B5b, 2026-10-15): 11,770 rows at PyAutoFit pin df6d218f10cd all admitted; accounting 11,750 scored (10,580 rows, 1,170 failed) + 20 probe + 250 deferred; all 20 per-realisation references complete (Nautilus-only), JAX blend realisation-1 rebuilt locally. Catalogue gaussian_x3@2 section orders legs within task and dataset (autofit_inference results/accounting, catalogue). Probe-first remainder (980 runs) is 409930 on ral; next: B5b-2 pulls and judges it; wave-2 summary records withheld on size pending autofit_inference#10.

+

Wave 2 closed and judged under gaussian_x3@2 (B5b-2, 2026-10-15): all 12,750 pulled rows admitted at PyAutoFit pin df6d218f10cd; of the 13,000 expected runs, 11,580 rows + 1,170 failed + 0 missing + 250 deferred. The 409930 probe-first legs close: NSS 50-step leads both evidence rankings (blend 158/250, separated 210/250); Zeus JAX is 0/250 on the blend and 190/250 on the separated control (first among separated posterior legs). The catalogue ranks only legs with success rate >= 50 % over >= 20 attempts and lists the rest as unreliable (post-hoc presentation convention, protocol §8 note; autofit_inference#12 ruling 2). Summary records published as inference-summary@3 (101 record shards). Next: after the B5b-2 PRs merge, capture the Insight fit snapshot from main under @3 and run pyauto-insight check online; then B5c (a second model family).

Runs

-
    -
  • 409930_[0-19] — open — ral — 2026-10-15 — wave-2 probe-first remainder (B5b): Zeus JAX + NSS n_live_200_mcmc_steps_50, both datasets, 980 runs after the probe under probe-derived @2 timeouts (NSS 600 s…
    • -
+

nothing on the cluster

Last 5

    +
  • 2026-10-15 — run — 409930_[0-19] finished — wall 1:30: wave-2 probe-first remainder (B5b): Zeus JAX + NSS n_live_200_mcmc_steps_50, both datasets, 980 runs after the probe under probe-derived @2 timeouts (NSS 600 s, Zeus 1168 s); %20; submit_wave2_remainder;…
  • 2026-10-15 — run — 408367_[0-389] finished — wall 1:57: scored wave 2 separated jax_cpu: 3750 runs, %60, afterany 408346:408347 — 390/390 COMPLETED; 3750 separated jax_cpu rows admitted; judged B5b
  • 2026-10-15 — run — 408366_[0-379] finished — wall 1:51: scored wave 2 separated numpy: 2250 runs, %60, afterany 408346:408347 — 380/380 COMPLETED; 2250 separated numpy rows admitted; judged B5b
  • 2026-10-15 — run — 408365_[0-344] finished — wall 1:46: scored wave 2 blend jax_cpu: 3500 runs, %60, afterany 408346:408347 — 345/345 COMPLETED; 3500 blend jax_cpu rows admitted; judged B5b
  • 2026-10-15 — run — 408348_[0-379] finished — wall 1:53: scored wave 2 (gaussian_x3@2) blend numpy: 2250 runs, 50 seeds x 5 realisations, %60, afterany 408346:408347; submit_wave2_scored_gaussian_x3_blend_numpy — 380/380 COMPLETED; 2250 blend numpy rows admit…
    • -
  • 2026-10-15 — run — 408347_[0-7] finished — wall 0:01: wave-2 separated reference builds, afterany 408245; %8 — 8/8 COMPLETED; separated reference builds

full log

resume autofit_inference

diff --git a/dashboard.md b/dashboard.md index 90d0317..75f88d2 100644 --- a/dashboard.md +++ b/dashboard.md @@ -11,7 +11,7 @@ One ledger per science project: what is on the cluster right now, what the human | Where | Count | |-------|------:| | [Running](#projects) | 22 | -| [Open](#projects) | 48 | +| [Open](#projects) | 47 | | [Projects](#projects) | 8 |
📋 check in since last time @@ -49,7 +49,7 @@ Last check-in: 2026-09-30T09:51Z. | slope_hierarchy_scale | 0 | 0 | 2026-10-14 | | ic50_workspace | 0 | 0 | 2026-10-14 | | autolens_inference | 6 | 0 | 2026-09-24 | -| autofit_inference | 0 | 1 | 2026-10-15 | +| autofit_inference | 0 | 0 | 2026-10-15 | ## Projects @@ -336,19 +336,19 @@ active · ral · [autofit_inference#8](https://github.com/PyAutoLabs/autofit_inf **Now** -Scored wave 2 pulled and judged under gaussian_x3@2 (B5b, 2026-10-15): 11,770 rows at PyAutoFit pin df6d218f10cd all admitted; accounting 11,750 scored (10,580 rows, 1,170 failed) + 20 probe + 250 deferred; all 20 per-realisation references complete (Nautilus-only), JAX blend realisation-1 rebuilt locally. Catalogue gaussian_x3@2 section orders legs within task and dataset (autofit_inference results/accounting, catalogue). Probe-first remainder (980 runs) is 409930 on ral; next: B5b-2 pulls and judges it; wave-2 summary records withheld on size pending autofit_inference#10. +Wave 2 closed and judged under gaussian_x3@2 (B5b-2, 2026-10-15): all 12,750 pulled rows admitted at PyAutoFit pin df6d218f10cd; of the 13,000 expected runs, 11,580 rows + 1,170 failed + 0 missing + 250 deferred. The 409930 probe-first legs close: NSS 50-step leads both evidence rankings (blend 158/250, separated 210/250); Zeus JAX is 0/250 on the blend and 190/250 on the separated control (first among separated posterior legs). The catalogue ranks only legs with success rate >= 50 % over >= 20 attempts and lists the rest as unreliable (post-hoc presentation convention, protocol §8 note; autofit_inference#12 ruling 2). Summary records published as inference-summary@3 (101 record shards). Next: after the B5b-2 PRs merge, capture the Insight fit snapshot from main under @3 and run pyauto-insight check online; then B5c (a second model family). **Runs** -- 409930_[0-19] — open — ral — 2026-10-15 — wave-2 probe-first remainder (B5b): Zeus JAX + NSS n_live_200_mcmc_steps_50, both datasets, 980 runs after the probe under probe-derived @2 timeouts (NSS 600 s… +- _nothing on the cluster_ **Last 5** +- 2026-10-15 — *run* — 409930_[0-19] finished — wall 1:30: wave-2 probe-first remainder (B5b): Zeus JAX + NSS n_live_200_mcmc_steps_50, both datasets, 980 runs after the probe under probe-derived @2 timeouts (NSS 600 s, Zeus 1168 s); %20; submit_wave2_remainder;… - 2026-10-15 — *run* — 408367_[0-389] finished — wall 1:57: scored wave 2 separated jax_cpu: 3750 runs, %60, afterany 408346:408347 — 390/390 COMPLETED; 3750 separated jax_cpu rows admitted; judged B5b - 2026-10-15 — *run* — 408366_[0-379] finished — wall 1:51: scored wave 2 separated numpy: 2250 runs, %60, afterany 408346:408347 — 380/380 COMPLETED; 2250 separated numpy rows admitted; judged B5b - 2026-10-15 — *run* — 408365_[0-344] finished — wall 1:46: scored wave 2 blend jax_cpu: 3500 runs, %60, afterany 408346:408347 — 345/345 COMPLETED; 3500 blend jax_cpu rows admitted; judged B5b - 2026-10-15 — *run* — 408348_[0-379] finished — wall 1:53: scored wave 2 (gaussian_x3@2) blend numpy: 2250 runs, 50 seeds x 5 realisations, %60, afterany 408346:408347; submit_wave2_scored_gaussian_x3_blend_numpy — 380/380 COMPLETED; 2250 blend numpy rows admit… -- 2026-10-15 — *run* — 408347_[0-7] finished — wall 0:01: wave-2 separated reference builds, afterany 408245; %8 — 8/8 COMPLETED; separated reference builds [full log](https://github.com/PyAutoLabs/PyAutoCortex/blob/main/projects/autofit_inference.md) diff --git a/projects/autofit_inference.md b/projects/autofit_inference.md index 209cf05..331caf4 100644 --- a/projects/autofit_inference.md +++ b/projects/autofit_inference.md @@ -5,14 +5,14 @@ Issue: autofit_inference#8 ## Now -Scored wave 2 pulled and judged under gaussian_x3@2 (B5b, 2026-10-15): 11,770 rows at PyAutoFit pin df6d218f10cd all admitted; accounting 11,750 scored (10,580 rows, 1,170 failed) + 20 probe + 250 deferred; all 20 per-realisation references complete (Nautilus-only), JAX blend realisation-1 rebuilt locally. Catalogue gaussian_x3@2 section orders legs within task and dataset (autofit_inference results/accounting, catalogue). Probe-first remainder (980 runs) is 409930 on ral; next: B5b-2 pulls and judges it; wave-2 summary records withheld on size pending autofit_inference#10. +Wave 2 closed and judged under gaussian_x3@2 (B5b-2, 2026-10-15): all 12,750 pulled rows admitted at PyAutoFit pin df6d218f10cd; of the 13,000 expected runs, 11,580 rows + 1,170 failed + 0 missing + 250 deferred. The 409930 probe-first legs close: NSS 50-step leads both evidence rankings (blend 158/250, separated 210/250); Zeus JAX is 0/250 on the blend and 190/250 on the separated control (first among separated posterior legs). The catalogue ranks only legs with success rate >= 50 % over >= 20 attempts and lists the rest as unreliable (post-hoc presentation convention, protocol §8 note; autofit_inference#12 ruling 2). Summary records published as inference-summary@3 (101 record shards). Next: after the B5b-2 PRs merge, capture the Insight fit snapshot from main under @3 and run pyauto-insight check online; then B5c (a second model family). ## Runs -- 409930_[0-19] — open — ral — 2026-10-15 — wave-2 probe-first remainder (B5b): Zeus JAX + NSS n_live_200_mcmc_steps_50, both datasets, 980 runs after the probe under probe-derived @2 timeouts (NSS 600 s, Zeus 1168 s); %20; submit_wave2_remainder; pin df6d218f10cd; pulled in B5b-2 ## Log +- 2026-10-15 — run — 409930_[0-19] finished — wall 1:30: wave-2 probe-first remainder (B5b): Zeus JAX + NSS n_live_200_mcmc_steps_50, both datasets, 980 runs after the probe under probe-derived @2 timeouts (NSS 600 s, Zeus 1168 s); %20; submit_wave2_remainder; pin df6d218f10cd; pulled in B5b-2 — 20/20 COMPLETED (16:08-17:38 UTC); 980 remainder rows pulled by per-directory tar (20 probe rows byte-identical); all admitted under gaussian_x3@2, 0 missing; judged B5b-2 - 2026-10-15 — run — 408367_[0-389] finished — wall 1:57: scored wave 2 separated jax_cpu: 3750 runs, %60, afterany 408346:408347 — 390/390 COMPLETED; 3750 separated jax_cpu rows admitted; judged B5b - 2026-10-15 — run — 408366_[0-379] finished — wall 1:51: scored wave 2 separated numpy: 2250 runs, %60, afterany 408346:408347 — 380/380 COMPLETED; 2250 separated numpy rows admitted; judged B5b - 2026-10-15 — run — 408365_[0-344] finished — wall 1:46: scored wave 2 blend jax_cpu: 3500 runs, %60, afterany 408346:408347 — 345/345 COMPLETED; 3500 blend jax_cpu rows admitted; judged B5b diff --git a/state.json b/state.json index 3167645..47ac776 100644 --- a/state.json +++ b/state.json @@ -4,7 +4,7 @@ "repo": "PyAutoCortex", "status": "yellow", "headline": "YELLOW \u2014 8 active \u00b7 22 running", - "updated": "2026-10-15T16:00:25Z", + "updated": "2026-10-15T17:38:42Z", "pages_url": "https://pyautolabs.github.io/PyAutoCortex/", "items": [ {