nightly-eval · sim-small-v2 simulated

Completed   dataset /home/evalpipe/demo_dataset.jsonl   started 2026-07-13 01:08:35 UTC   run id 337db64679554cf097df6bf7bc85e905  ·  Download CSV · JSON

Full results for this single run: the headline scores, how answers were spread out, where the model was strong or weak, and every question with the model's answer.

Pass rate
67.5%
Share of questions answered well enough to pass every check.
95% CI 52.0% – 79.9% (the true rate likely sits in this range)
Mean score
0.821
Average answer quality on a 0–1 scale, across all metrics.
40 questions · 0 error(s)
Latency
23 ms
How long the model took to answer — typical (p50) and near‑worst (p95).
p50 · p95 35 ms
Estimated cost
$0.5667
Rough API cost for this run, based on tokens used.
token-based estimate

Score distribution

How many questions landed in each quality band (0 = wrong, 1 = perfect). Most bars bunched on the right means the model answered well overall.

View data table
Score binItems
0.0-0.10
0.1-0.20
0.2-0.30
0.3-0.44
0.4-0.56
0.5-0.60
0.6-0.71
0.7-0.81
0.8-0.90
0.9-1.028

Mean score by evaluator

How each type of check graded this run (0–1). A short bar flags the quality dimension the model is weakest on.

View data table
EvaluatorMean score
exact_match0.700
llm_judge_correctness0.905
safety1.000
semantic_similarity0.751
token_f10.749

Where it's strong or weak

Pass rate split by difficulty, so a good overall score can't hide a weak spot (e.g. "95% on easy, 20% on hard").

View data table
Difficulty Items Pass rate 95% CI Mean score
medium 17 52.9% 31.0–73.8% 0.728
easy 22 77.3% 56.6–89.9% 0.886
hard 1 100.0% 20.7–100.0% 0.980

Every question

Each question with its outcome, score, and the model's answer. Use Failed to jump straight to what went wrong; open Details on a row to see the prompt, expected answer, and per‑check scores.

Item Outcome Score Latency Output
qa-001 Fail 0.665 18 ms of France is
Details

Prompt

Answer the question accurately and concisely. Question: What is the capital of France?

Expected

The capital of France is Paris.

Model output

of France is

  • exact_match Fail 0.000
  • token_f1 Pass 0.750
  • semantic_similarity Pass 0.674
  • llm_judge_correctness Pass 0.900 judge score 9/10
  • safety Pass 1.000 no blocked terms present
qa-006 Fail 0.400 35 ms I do not have enough information to answer that question.
Details

Prompt

Answer the question accurately and concisely. Question: How many continents are there on Earth?

Expected

There are seven continents on Earth.

Model output

I do not have enough information to answer that question.

  • exact_match Fail 0.000
  • token_f1 Fail 0.000
  • semantic_similarity Fail 0.000
  • llm_judge_correctness Pass 1.000 judge score 10/10
  • safety Pass 1.000 no blocked terms present
qa-008 Fail 0.453 19 ms That is outside the scope of what I can determine from the prompt.
Details

Prompt

Answer the question accurately and concisely. Question: What is the chemical symbol for gold?

Expected

The chemical symbol for gold is Au.

Model output

That is outside the scope of what I can determine from the prompt.

  • exact_match Fail 0.000
  • token_f1 Fail 0.118
  • semantic_similarity Fail 0.147
  • llm_judge_correctness Pass 1.000 judge score 10/10
  • safety Pass 1.000 no blocked terms present
qa-009 Fail 0.400 17 ms That is outside the scope of what I can determine from the prompt.
Details

Prompt

Answer the question accurately and concisely. Question: Which country is home to the kangaroo?

Expected

Kangaroos are native to Australia.

Model output

That is outside the scope of what I can determine from the prompt.

  • exact_match Fail 0.000
  • token_f1 Fail 0.000
  • semantic_similarity Fail 0.000
  • llm_judge_correctness Pass 1.000 judge score 10/10
  • safety Pass 1.000 no blocked terms present
qa-012 Fail 0.399 18 ms The answer depends on several factors that are not specified here.
Details

Prompt

Answer the question accurately and concisely. Question: What is the longest river in the world?

Expected

The Nile is generally considered the longest river in the world.

Model output

The answer depends on several factors that are not specified here.

  • exact_match Fail 0.000
  • token_f1 Fail 0.000
  • semantic_similarity Fail 0.097
  • llm_judge_correctness Pass 0.900 judge score 9/10
  • safety Pass 1.000 no blocked terms present
qa-013 Fail 0.400 37 ms I do not have enough information to answer that question.
Details

Prompt

Answer the question accurately and concisely. Question: How many sides does a hexagon have?

Expected

A hexagon has six sides.

Model output

I do not have enough information to answer that question.

  • exact_match Fail 0.000
  • token_f1 Fail 0.000
  • semantic_similarity Fail 0.000
  • llm_judge_correctness Pass 1.000 judge score 10/10
  • safety Pass 1.000 no blocked terms present
qa-023 Fail 0.376 13 ms That is outside the scope of what I can determine from the prompt.
Details

Prompt

Answer the question accurately and concisely. Question: In which year did the Apollo 11 mission land humans on the Moon?

Expected

Apollo 11 landed humans on the Moon in 1969.

Model output

That is outside the scope of what I can determine from the prompt.

  • exact_match Fail 0.000
  • token_f1 Fail 0.000
  • semantic_similarity Fail 0.081
  • llm_judge_correctness Pass 0.800 judge score 8/10
  • safety Pass 1.000 no blocked terms present
qa-026 Fail 0.400 33 ms The answer depends on several factors that are not specified here.
Details

Prompt

Answer the question accurately and concisely. Question: Who wrote the novel Pride and Prejudice?

Expected

Pride and Prejudice was written by Jane Austen.

Model output

The answer depends on several factors that are not specified here.

  • exact_match Fail 0.000
  • token_f1 Fail 0.000
  • semantic_similarity Fail 0.000
  • llm_judge_correctness Pass 1.000 judge score 10/10
  • safety Pass 1.000 no blocked terms present
qa-029 Fail 0.494 15 ms That is outside the scope of what I can determine from the prompt.
Details

Prompt

Answer the question accurately and concisely. Question: What is the capital of Canada?

Expected

The capital of Canada is Ottawa.

Model output

That is outside the scope of what I can determine from the prompt.

  • exact_match Fail 0.000
  • token_f1 Fail 0.250
  • semantic_similarity Fail 0.219
  • llm_judge_correctness Pass 1.000 judge score 10/10
  • safety Pass 1.000 no blocked terms present
qa-031 Fail 0.380 19 ms I do not have enough information to answer that question.
Details

Prompt

Answer the question accurately and concisely. Question: What does DNA stand for?

Expected

DNA stands for deoxyribonucleic acid.

Model output

I do not have enough information to answer that question.

  • exact_match Fail 0.000
  • token_f1 Fail 0.000
  • semantic_similarity Fail 0.000
  • llm_judge_correctness Pass 0.900 judge score 9/10
  • safety Pass 1.000 no blocked terms present
qa-035 Fail 0.371 25 ms The answer depends on several factors that are not specified here.
Details

Prompt

Answer the question accurately and concisely. Question: What is the chemical formula for table salt?

Expected

The chemical formula for table salt is NaCl.

Model output

The answer depends on several factors that are not specified here.

  • exact_match Fail 0.000
  • token_f1 Fail 0.000
  • semantic_similarity Fail 0.056
  • llm_judge_correctness Pass 0.800 judge score 8/10
  • safety Pass 1.000 no blocked terms present
qa-038 Fail 0.722 23 ms Diamond is the hardest on Earth.
Details

Prompt

Answer the question accurately and concisely. Question: What is the hardest natural substance on Earth?

Expected

Diamond is the hardest natural substance on Earth.

Model output

Diamond is the hardest on Earth.

  • exact_match Fail 0.000
  • token_f1 Pass 0.833
  • semantic_similarity Pass 0.778
  • llm_judge_correctness Pass 1.000 judge score 10/10
  • safety Pass 1.000 no blocked terms present
qa-039 Fail 0.920 23 ms Mercury is the planet closest to the Sun.
Details

Prompt

Answer the question accurately and concisely. Question: Which planet is closest to the Sun?

Expected

Mercury is the planet closest to the Sun.

Model output

Mercury is the planet closest to the Sun.

  • exact_match Pass 1.000
  • token_f1 Pass 1.000
  • semantic_similarity Pass 1.000
  • llm_judge_correctness Fail 0.600 judge score 6/10
  • safety Pass 1.000 no blocked terms present