wassnameandClaudypoo 54b8a2bb79 Add real Judgemark v4 frontier and evaluator-blind-spot papers ...

Replace the hand-waved judge-model guidance with the checked-in Judgemark v4
cost-vs-score frontier (opus-4-6 top, gemma-4-31b the cheap knee; Qwen/DeepSeek
are mid-pack not frontier, correcting the earlier claim). Quote Doddapaneni
(evaluators miss >50% of injected quality drops) and Shankar (criteria drift,
validators need validation). Cache both papers and the frontier table.

Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>

2026-07-22 16:12:46 +08:00

..

checklist.md

docs: resolve ml-debug TODO references

2026-06-12 06:52:38 +08:00

diagnostics.md

Correct overconfident debugging advice

2026-07-12 22:00:49 +08:00

llm_judges.md

Add real Judgemark v4 frontier and evaluator-blind-spot papers

2026-07-22 16:12:46 +08:00

loss_surface.md

Update loss_surface.md

2026-06-11 21:21:01 +08:00

metric_stuck.md

restructure: folklore-first, quote-verified, with wassname intro

2026-06-02 20:46:25 +08:00

research_taste.md

Cite Irpan in research taste (signs-of-life, seed canary)

2026-06-25 10:31:39 +08:00

static_analysis.md

feat(ml_debug): add JAX grep patterns and diagnostic equivalents

2026-03-06 14:10:39 +08:00

sweeps.md

restructure: folklore-first, quote-verified, with wassname intro

2026-06-02 20:46:25 +08:00

transformers.md

docs: add sourced transformer report folklore

2026-06-12 07:02:58 +08:00