Spanish AEPD corpus test
X post comparing Jev with a hand-built regex on 544 Spanish data-protection resolutions: 98.2% agreement for about five cents.
This summary is based on how curated lists describe the project; its README was not available to review.
Overview
This is a post on X. According to the directory summary, the author compared Jev with a hand-built regular expression on 544 public resolutions from Spain's data-protection authority (AEPD), finding 98.2% agreement for a total cost of about five cents.
No post text was fetched, so the exact field being extracted or classified and the nature of the disagreements are unknown here. It is a practical data point that, for a task already solvable by a careful regex, Jev reached near-identical results with no hand-written rules at negligible cost.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Other | The classification the regex encoded, asked of each resolution(inferred) | — |
How it works
- Collect 544 public AEPD resolutions
- Run the hand-built regex and Jev on each
- Compare agreement and cost
Strengths
- 98.2% agreement with a regex for about five cents, per the summary
Caveats
- Grounded only in a one-line summary
- Agreement with a regex is not accuracy against ground truth
At a glance
- Type
- Discussion
- Status
- discussion
Use cases and integrations
How curated lists describe it
Jev versus a hand-built regex on 544 public data-protection resolutions: 98.2% agreement for about five cents.
hellogumbo/awesome-jev
Frequently asked questions
What was tested?
Jev versus a hand-built regex on 544 Spanish data-protection resolutions.
What was the result?
98.2% agreement for about five cents, per the summary.
Where is it?
x.com/juanmacias/status/2100463494629925048.
Related projects
Anthony Maio's Substack essay on what a model that cannot generate text is for, with a critical look at Jev's claims.
ArticlediscussionHacker News thread on Decision Index 0.2.1, which compared about 70 open reproductions of Jev's decision model.
DiscussiondiscussionVideo post on X comparing Jev with a structured-output Qwen baseline running on Cerebras.
DiscussiondiscussionOpen-source BYOK arena where Jev faces another model as a judge on your question, with blind voting before costs are revealed.
ProjectbetaTypeScriptNovelty demo that turns Jev answers to 22 Boolean gate questions into lookup tables for a simulated RISC-V CPU running compiled C.
ProjectprototypeBenchmark of Jev versus rules, local PII models and LLMs at finding personal information: accuracy, leaks, cost and latency.
BenchmarkresearchPython