SeterusnyaPanduan seterusnya
LLM Eval Frameworks: Promptfoo and DeepEval
Teknikal
PANDUAN Teknikal
An LLM eval dataset is a curated set of realistic inputs, each paired with a golden answer, reference facts or a grading rubric, that you run your application against to measure quality and catch regressions.
Dengar panduan iniAI-diceritakan · ~3 min
It matters because a change to the prompt, model or retrieval can improve one behavior while breaking another. Without a fixed test set, teams end up judging changes by a few hand-picked examples.
Start with real usage. Logs, support tickets, search queries and user interviews show what people actually ask, including the messy phrasing that invented examples miss. Before launch, have domain experts write inputs. You can add synthetic cases generated by an LLM, as long as a person reviews them and removes unrealistic or duplicate items. Next, define what correct means for each case. Some tasks have one right answer, such as classification, extraction or a factual lookup. For these, store a golden answer that can be compared exactly or after light normalization. For open-ended tasks, write a rubric of specific, checkable criteria, such as 'mentions the refund deadline' or 'does not recommend a competitor'. Vague criteria like 'is helpful' lead to inconsistent grading, whether the grader is a person or a judge model. Then cover the range of real inputs on purpose. Tag cases by intent, difficulty, language and user type, and check that the important groups are represented. Add edge cases and adversarial inputs: - ambiguous questions - requests with missing information - very long inputs - out-of-scope requests that should be declined - prompt-injection attempts Keep a regression slice made of past failures. Size the set to the decision it has to support. A few dozen well-chosen cases can reveal large problems early. Detecting small differences reliably takes more. With 100 pass/fail cases and a pass rate near 80 percent, the margin of error is roughly plus or minus 8 percentage points at 95 percent confidence. A two-point improvement is indistinguishable from noise at that size. Report results for each slice, not only an overall average. A common mistake is to treat the dataset as finished. Products and users change, so refresh it from new logs. Keep a held-out portion that you never tune against, so your prompts do not overfit to the test.
A support team exports 300 real tickets with personal data removed, labels the correct resolution for each, and scores every new prompt version against them.
For a meeting summarizer where many wordings are correct, the team writes a rubric: covers the three key decisions, makes no unsupported claims, stays under 150 words.
The team adds edge cases: an empty input, a question in Spanish, a document containing a prompt-injection attempt, and an out-of-scope request the assistant should decline.
When a user reports a bad answer in production, the case goes into the dataset with the correct answer, so the fix stays tested from then on.
Eval tooling has matured quickly. Open-source frameworks and hosted platforms now manage datasets, run graders and track results over time. Synthetic data generation will probably keep improving, which makes rare cases cheaper to cover. It cannot replace real user inputs as the anchor for what matters. As applications become more agentic, eval sets increasingly include multi-step tasks graded on both the final outcome and the steps taken along the way. The core discipline changes slowly: define success clearly, cover real inputs, use enough cases to trust the numbers, and refresh the set regularly from production.
Newsletter percuma
Tiga cerita AI yang disahkan setiap pagi hari bekerja, ditulis dalam Bahasa Inggeris biasa. Percuma selamanya, tiada iklan.
Satu e-mel setiap hari bekerja. Berhenti melanggan dalam satu klik. Kami tidak pernah menjual atau berkongsi alamat anda.
Uji diri anda
Maklum balas segera pada setiap jawapan, dan sijil yang boleh dikongsi dengan ID yang boleh disahkan sebaik sahaja anda lulus kursus.
Sokong pendidikan AI percuma. AI Understanding ialah pertubuhan bukan keuntungan 501(c)(3) — tiada iklan, tiada bayaran dinding, selamanya. Buat derma
An LLM eval dataset is a curated set of realistic inputs, each paired with a golden answer, reference facts or a grading rubric, that you run your application against to measure quality and catch regressions. It matters because a change to the prompt, model or retrieval can improve one behavior while breaking another. Without a fixed test set, teams end up judging changes by a few hand-picked examples.
Penggunaan sebenar menangkap perkara yang sebenarnya orang tanya, termasuk frasa yang tidak kemas. Kes sintetik adalah suplemen yang disemak, bukan pengganti.
Output terbuka seperti ringkasan tidak boleh dipadankan dengan tepat. Kriteria yang khusus dan boleh disemak membolehkan penggred menilai mereka secara konsisten.
Kriteria yang baik adalah khusus dan boleh disemak. Kriteria yang tidak jelas seperti 'membantu' membawa kepada penggredan yang tidak konsisten oleh orang dan model hakim.
Pada keyakinan 95 peratus margin adalah kira-kira 8 mata pada saiz ini. Perbezaan kecil antara versi tidak boleh dipercayai tanpa lebih banyak kes.
Menambah setiap kegagalan tetap pada set data bermakna perubahan kemudian tidak dapat memperkenalkan semula pepijat yang sama secara senyap-senyap.
Teruskan belajar
Lebih banyak panduan dipilih untuk topik ini
SeterusnyaPanduan seterusnya
LLM Eval Frameworks: Promptfoo and DeepEval
Teknikal