DietrichGebertandClaude Opus 4.8
caf138df56
* fix(benchmarks): correctness gate scores unfenced code; fix debounce task The `correct` gate under-reported correctness for terse models, the likely source of "Ponytail degrades models" reports (issue #65): - extractBlocks() only matched fenced code blocks, so bare/unfenced code scored an automatic fail even when correct. Now falls back to the whole response as one block (and tolerates CRLF). Debounce detection also accepts unfenced arrow functions. - The debounce task asked to "add debounce to a search input" but the check expected a reusable debounce(fn, delay) util, failing correct inline answers. Task reworded to the deliverable the check verifies. Adds correctness.test.js (regression guard) and a GPT-mini repro config plus results writeup: on a clean n=20 run, the reported gpt-4.1-mini drop (10/15) does not reproduce (100/100). The LOC win (~halved) holds. README repro fixed: promptfoo needs --env-file ../.env (reads cwd, not root). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(benchmarks): add robustness audit — ponytail vs baseline on edge cases Answers the real question behind #65: does ponytail's push for the shortest solution make weak models produce wrong code on edge cases? robustness-audit.js: 16 self-verifying tasks (12 algorithmic edge-case traps + 4 validators). Each check ships a known-good and known-lazy-wrong reference that must pass/fail before any model output is scored (--selftest, 16/16). Findings (gpt-4.1-mini + gpt-5.4-mini, baseline vs ponytail): parity on every edge-case trap on both models. The one measured soft spot is gpt-5.4-mini email (~4-5%, reaches for parseaddr). A sharpened SKILL.md validation rule had no reliable effect in an n=100 A/B (96% vs 95%), so it was not shipped — the tendency is model-level, not skill-level. Full writeup in results/. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * test(benchmarks): email slip is provider-specific — 100% on Claude High-n cross-provider follow-up to the robustness audit. The one ponytail soft spot (email validation via parseaddr) splits by provider, not model size: - Claude (haiku/sonnet/opus): 100% under ponytail, n=40 each — and ponytail beats baseline (unconstrained Sonnet over-engineers into an always-truthy dict, 0/40; ponytail writes a clean validator). - OpenAI (gpt-4.1-mini..gpt-5.5): slips at every size under ponytail (~79-98%), baseline ~100%. The parseaddr reflex lives in OpenAI training. Not fixable by skill text: 8 distinct SKILL.md edits (incl. an n=100 A/B, 96% vs 95%) all scored <= current, several worse, all bloated LOC. Nothing shipped. SKILL.md unchanged. Conclusion: on ponytail's target platform (Claude) email is 100%; the GPT slip is a documented cross-provider transfer quirk. Adds model-email.js / claude-email.js to reproduce the tables. Writeup updated. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * docs(benchmarks): correct misleading Sonnet baseline 0 percent The Sonnet baseline 0/40 on email is a return-type artifact, not a logic failure: unconstrained Sonnet returns a dict {is_valid, message} instead of a bool, so the bool-contract gate scores every case as accepted. Read dict-aware via is_valid, its logic is ~75% correct (9/12). Reframed honestly so we are not presenting 0 vs 100 as a clean win; ponytail still wins (clean 100% bool) but the point is over-engineered return type, not total failure. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
40 lines
1.9 KiB
JavaScript
40 lines
1.9 KiB
JavaScript
// Cross-model email rate at high n: is the parseaddr quirk gpt-5.4-mini-specific?
|
|
const fs = require('fs'), path = require('path');
|
|
const { checkPy, pyBlock, TASKS } = require('./robustness-audit.js');
|
|
const skill = fs.readFileSync(path.join(__dirname, '..', 'skills', 'ponytail', 'SKILL.md'), 'utf8');
|
|
const email = TASKS.find(t => t.name === 'email');
|
|
const N = Number(process.env.ME_N) || 100;
|
|
const MODELS = (process.env.ME_MODELS || 'gpt-4.1-mini,gpt-5.4-mini').split(',');
|
|
|
|
const kv = Object.fromEntries(fs.readFileSync(path.join(__dirname, '..', '.env'), 'utf8')
|
|
.split(/\r?\n/).filter(l => l.includes('=') && !l.trim().startsWith('#'))
|
|
.map(l => { const i = l.indexOf('='); return [l.slice(0, i).trim(), l.slice(i + 1).trim()]; }));
|
|
const KEY = kv.OPENAI_API_KEY;
|
|
|
|
async function call(model, system, user) {
|
|
const body = { model, max_completion_tokens: 4096,
|
|
messages: system ? [{ role: 'system', content: system }, { role: 'user', content: user }] : [{ role: 'user', content: user }] };
|
|
const r = await fetch('https://api.openai.com/v1/chat/completions', { method: 'POST',
|
|
headers: { Authorization: 'Bearer ' + KEY, 'Content-Type': 'application/json' }, body: JSON.stringify(body) });
|
|
if (!r.ok) return { err: r.status };
|
|
return { text: (await r.json()).choices?.[0]?.message?.content || '' };
|
|
}
|
|
|
|
(async () => {
|
|
console.log(`email, n=${N}\n`);
|
|
console.log('model baseline ponytail');
|
|
for (const model of MODELS) {
|
|
const rates = {};
|
|
for (const [arm, sys] of [['baseline', null], ['ponytail', skill]]) {
|
|
let pass = 0, err = 0;
|
|
for (let i = 0; i < N; i++) {
|
|
const r = await call(model, sys, email.prompt);
|
|
if (r.err) { err++; continue; }
|
|
if (checkPy(pyBlock(r.text), email)) pass++;
|
|
}
|
|
rates[arm] = `${pass}/${N - err}`;
|
|
}
|
|
console.log(`${model.padEnd(15)} ${rates.baseline.padEnd(10)} ${rates.ponytail}`);
|
|
}
|
|
})();
|