wassname

wassname synced and deleted reference refs/heads/add-bullshit-benchmark-loader at wassname/persona-steering-template-library from mirror 2026-08-09 12:20:59 +08:00

wassname synced commits to refs/heads/main at wassname/persona-steering-template-library from mirror 2026-08-09 12:20:59 +08:00

5e177a2745 Merge pull request #3 from wassname/add-bullshit-benchmark-loader

8353897780 scenarios: register BullshitBench for credulity screening

0dde39e8d0 test: cover BullshitBench scenario loading

9639ae3c81 fix: repair scenario exporter paths after script move

0b53320bdf scenarios: add the bullshit benchmark (55 incoherent-premise questions)

Compare 5 commits »

wassname synced commits to refs/heads/main at wassname/ml-debug from mirror 2026-08-09 12:10:49 +08:00

996942c4de Make ML debugging expose evidence and uncertainty

wassname synced commits to refs/heads/main at wassname/ml_debug from mirror 2026-08-09 12:10:43 +08:00

996942c4de Make ML debugging expose evidence and uncertainty

wassname synced commits to refs/heads/main at wassname/ml-debug from mirror 2026-08-08 11:22:26 +08:00

4e45bd4140 DPO can barely move: tinker's own reference run reports 0.52 pair accuracy

b5a70354a4 Judgemark: add the cheap-tier rows and date the snapshot

602f6193ed assume every negative result is a bug until the logs rule it out; be patient

Compare 3 commits »

wassname synced commits to refs/heads/main at wassname/ml_debug from mirror 2026-08-08 11:22:26 +08:00

4e45bd4140 DPO can barely move: tinker's own reference run reports 0.52 pair accuracy

b5a70354a4 Judgemark: add the cheap-tier rows and date the snapshot

602f6193ed assume every negative result is a bug until the logs rule it out; be patient

Compare 3 commits »

wassname synced commits to refs/heads/main at wassname/pi-goals from mirror 2026-08-05 13:20:49 +08:00

751e20b7ee drafting prompt: this skeleton wins over any plan format a loaded skill also supplies

1cd6e04e05 README: the fold, user voice, learnings, unlimited appendix; the judge never runs verify

07510d96bd fold the plan at ## Log: re-send the working set on a stale cadence, the whole file only on resync

Compare 3 commits »

wassname synced commits to refs/heads/main at wassname/pi-plan from mirror 2026-08-05 13:20:28 +08:00

751e20b7ee drafting prompt: this skeleton wins over any plan format a loaded skill also supplies

1cd6e04e05 README: the fold, user voice, learnings, unlimited appendix; the judge never runs verify

07510d96bd fold the plan at ## Log: re-send the working set on a stale cadence, the whole file only on resync

Compare 3 commits »

wassname synced commits to refs/heads/master at wassname/Unsupervised-Elicitation from mirror 2026-08-05 12:50:34 +08:00

175641a541 readme: drop the count word before intrinsic/extrinsic

8e394472df readme: quote the model's own words, intrinsic not implicit, drop dangling 'above'

48c9214d9f readme: expand ICM, commit the plot images, replace pasted upstream readme with a link

fc6f5896a8 readme: lead with what concept ICM picked, fold in LW comment, collapse upstream readme

Compare 4 commits »

wassname synced commits to refs/heads/backprop at wassname/Unsupervised-Elicitation from mirror 2026-08-05 12:50:33 +08:00

ffa2c9b808 readme: drop the count word before intrinsic/extrinsic

ba91650af6 readme: quote the model's own words, intrinsic not implicit, drop dangling 'above'

05150136d9 readme: expand ICM, commit the plot images, replace pasted upstream readme with a link

86d88f3d40 readme: lead with what concept ICM picked, fold in LW comment, collapse upstream readme

Compare 4 commits »

wassname synced commits to refs/heads/main at wassname/wassname from mirror 2026-08-03 13:20:35 +08:00

20d1e8ce63 Drop the motive item and fix the w2schar image URL

008ad6e8bd Retag the intro, reorder current work, shrink the images

b735ad6175 Update README.md

Compare 3 commits »

wassname synced commits to refs/heads/main at wassname/wassname from mirror 2026-07-31 12:50:59 +08:00

418d8fff75 Feature Moral Maps and restore the plot guide

8f5bc4928c Keep the profile copy reader-facing

5cbd9ecf38 Clarify the steering result is out of sample

83614852c0 Name the steering benchmark directly

c81fc9b2f6 Use the thread wording for current research

Compare 5 commits »

wassname synced commits to refs/heads/main at wassname/machiavelli_deep_value from mirror 2026-07-31 12:31:01 +08:00

025e2e6eb6 Compare only open Qwen 3.5 models

aebb86ce25 Focus the motive headline on Qwen models

fe2e63b6fa Add the no-rubric motive scaling figure

b161f7c060 docs: update README for six-model release (1680 records, $15.45 run cost)

Compare 4 commits »

wassname synced commits to refs/heads/main at wassname/machiavelli_deep_value from mirror 2026-07-30 12:20:46 +08:00

2dc71c67cf Retry transient generation rate limits

wassname synced commits to refs/heads/main at wassname/ml_debug from mirror 2026-07-26 13:27:45 +08:00

c7f22b8478 Add deployment-faithful time-series evaluation guide

a9b154fc6d add FPP3 (Hyndman & Athanasopoulos) scraped to markdown, 13 chapters + appendices

4fcea49ebb Show explicit pandas failure-score aggregation

0cc01db100 Clarify NaN selection bias in judge comparisons

5dd18385bd Fail judge aggregation on missing paired scores

Compare 5 commits »

wassname synced new reference refs/heads/add-bullshit-benchmark-loader to wassname/persona-steering-template-library from mirror 2026-07-26 13:27:37 +08:00

wassname synced commits to refs/heads/add-bullshit-benchmark-loader at wassname/persona-steering-template-library from mirror 2026-07-26 13:27:37 +08:00

wassname synced commits to refs/heads/main at wassname/ml-debug from mirror 2026-07-26 13:27:35 +08:00

c7f22b8478 Add deployment-faithful time-series evaluation guide

a9b154fc6d add FPP3 (Hyndman & Athanasopoulos) scraped to markdown, 13 chapters + appendices

4fcea49ebb Show explicit pandas failure-score aggregation

0cc01db100 Clarify NaN selection bias in judge comparisons

5dd18385bd Fail judge aggregation on missing paired scores

Compare 5 commits »

wassname synced commits to refs/heads/main at wassname/persona-steering-template-library from mirror 2026-07-25 13:20:56 +08:00

94d839629c Update trying_mean_mass_diff.md

df4979f926 Update trying_mean_mass_diff.md

680fad5a31 Create trying_mean_mass_diff.md

77e739a3cc docs(skill): rename scenario 'pool' to 'dataset', annotate Links with when-to-open

7ea078f520 docs(skill): surface scenario pools from SKILL.md entry point

Compare 5 commits »

wassname synced commits to refs/heads/main at wassname/ml_debug from mirror 2026-07-25 13:20:38 +08:00

38ddb544d9 Make 'Numbers worth knowing' quote-anchored with source-credence + calibrated language

3f35addb7f Expand 'Numbers worth knowing' bullets with point + context for readability

23d3331ee2 Fix two varglite nits: complete Lost-in-Middle quote, add context to config block

f10151e661 Add varglite 2026 judge lit-review + operational-numbers digest

3b7e82d402 Add 2026 lit-search batch to judge-bias evidence (position/self/reasoning/context-rot + benchmark index)

Compare 5 commits »