Hugging Face's logo

Datasets:
metrum-ai
/
prompt-library

Tasks:
Text Generation
Modalities:
Tabular
Text
Formats:
parquet
Size:
100K - 1M
Tags:
benchmarking
inference
load-testing
prompts
Libraries:
Datasets
Dask
Polars
License:
Dataset card Data Studio Files
xet
 Community
1
Dataset Viewer
Auto-converted to Parquet Duplicate
prompt
stringlengths
59
44.1k
prompt_length
int64
18
6.36k
target_output_length
int64
18
1.86k
reasoning
bool
2 classes
target_input_tokens
int64
32
7k
input_token_bucket
int64
32
7k
target_output_tokens
int64
32
2.05k
output_token_bucket
int64
32
2.05k
actual_words
int64
11
6.16k
A band has five members. If each member plays two instruments, how many instruments does the band play in total?
24
22
true
32
32
32
32
20
A band has five members. If each member plays two instruments, how many instruments does the band play in total?
24
30
true
32
32
48
48
20
A band has five members. If each member plays two instruments, how many instruments does the band play in total?
24
53
true
32
32
64
64
20
A band has five members. If each member plays two instruments, how many instruments does the band play in total?
24
92
true
32
32
128
128
20
A band has five members. If each member plays two instruments, how many instruments does the band play in total?
24
169
true
32
32
192
192
20
A band has five members. If each member plays two instruments, how many instruments does the band play in total?
24
185
true
32
32
256
256
20
A band has five members. If each member plays two instruments, how many instruments does the band play in total?
24
267
true
32
32
384
384
20
A band has five members. If each member plays two instruments, how many instruments does the band play in total?
24
348
true
32
32
512
512
20
A band has five members. If each member plays two instruments, how many instruments does the band play in total?
24
536
true
32
32
768
768
20
A band has five members. If each member plays two instruments, how many instruments does the band play in total?
24
679
true
32
32
1,024
1,024
20
A band has five members. If each member plays two instruments, how many instruments does the band play in total?
24
1,046
true
32
32
1,536
1,536
20
A band has five members. If each member plays two instruments, how many instruments does the band play in total?
24
1,521
true
32
32
2,048
2,048
20
What is the main idea behind Immanuel Kant's categorical imperative in moral philosophy and ethical reasoning?
22
28
true
32
32
32
32
16
What is the main idea behind Immanuel Kant's categorical imperative in moral philosophy and ethical reasoning?
22
32
true
32
32
48
48
16
What is the main idea behind Immanuel Kant's categorical imperative in moral philosophy and ethical reasoning?
22
44
true
32
32
64
64
16
What is the main idea behind Immanuel Kant's categorical imperative in moral philosophy and ethical reasoning?
22
93
true
32
32
128
128
16
What is the main idea behind Immanuel Kant's categorical imperative in moral philosophy and ethical reasoning?
22
143
true
32
32
192
192
16
What is the main idea behind Immanuel Kant's categorical imperative in moral philosophy and ethical reasoning?
22
172
true
32
32
256
256
16
What is the main idea behind Immanuel Kant's categorical imperative in moral philosophy and ethical reasoning?
22
247
true
32
32
384
384
16
What is the main idea behind Immanuel Kant's categorical imperative in moral philosophy and ethical reasoning?
22
393
true
32
32
512
512
16
What is the main idea behind Immanuel Kant's categorical imperative in moral philosophy and ethical reasoning?
22
635
true
32
32
768
768
16
What is the main idea behind Immanuel Kant's categorical imperative in moral philosophy and ethical reasoning?
22
712
true
32
32
1,024
1,024
16
What is the main idea behind Immanuel Kant's categorical imperative in moral philosophy and ethical reasoning?
22
1,132
true
32
32
1,536
1,536
16
What is the main idea behind Immanuel Kant's categorical imperative in moral philosophy and ethical reasoning?
22
1,618
true
32
32
2,048
2,048
16
In a small town, there are three social clubs. Each club has twenty members. Analyze interactions.
21
21
true
32
32
32
32
16
In a small town, there are three social clubs. Each club has twenty members. Analyze interactions.
21
38
true
32
32
48
48
16
In a small town, there are three social clubs. Each club has twenty members. Analyze interactions.
21
44
true
32
32
64
64
16
In a small town, there are three social clubs. Each club has twenty members. Analyze interactions.
21
91
true
32
32
128
128
16
In a small town, there are three social clubs. Each club has twenty members. Analyze interactions.
21
170
true
32
32
192
192
16
In a small town, there are three social clubs. Each club has twenty members. Analyze interactions.
21
177
true
32
32
256
256
16
In a small town, there are three social clubs. Each club has twenty members. Analyze interactions.
21
331
true
32
32
384
384
16
In a small town, there are three social clubs. Each club has twenty members. Analyze interactions.
21
362
true
32
32
512
512
16
In a small town, there are three social clubs. Each club has twenty members. Analyze interactions.
21
611
true
32
32
768
768
16
In a small town, there are three social clubs. Each club has twenty members. Analyze interactions.
21
773
true
32
32
1,024
1,024
16
In a small town, there are three social clubs. Each club has twenty members. Analyze interactions.
21
1,099
true
32
32
1,536
1,536
16
In a small town, there are three social clubs. Each club has twenty members. Analyze interactions.
21
1,221
true
32
32
2,048
2,048
16
Identify the continent where Egypt is located and describe its significance in world history.
21
21
true
32
32
32
32
14
Identify the continent where Egypt is located and describe its significance in world history.
21
30
true
32
32
48
48
14
Identify the continent where Egypt is located and describe its significance in world history.
21
48
true
32
32
64
64
14
Identify the continent where Egypt is located and describe its significance in world history.
21
87
true
32
32
128
128
14
Identify the continent where Egypt is located and describe its significance in world history.
21
132
true
32
32
192
192
14
Identify the continent where Egypt is located and describe its significance in world history.
21
164
true
32
32
256
256
14
Identify the continent where Egypt is located and describe its significance in world history.
21
232
true
32
32
384
384
14
Identify the continent where Egypt is located and describe its significance in world history.
21
451
true
32
32
512
512
14
Identify the continent where Egypt is located and describe its significance in world history.
21
536
true
32
32
768
768
14
Identify the continent where Egypt is located and describe its significance in world history.
21
802
true
32
32
1,024
1,024
14
Identify the continent where Egypt is located and describe its significance in world history.
21
1,100
true
32
32
1,536
1,536
14
Identify the continent where Egypt is located and describe its significance in world history.
21
1,333
true
32
32
2,048
2,048
14
An archaeologist discovered five ancient artifacts at a site. Each artifact dates back 2000 years.
22
22
true
32
32
32
32
15
An archaeologist discovered five ancient artifacts at a site. Each artifact dates back 2000 years.
22
28
true
32
32
48
48
15
An archaeologist discovered five ancient artifacts at a site. Each artifact dates back 2000 years.
22
39
true
32
32
64
64
15
An archaeologist discovered five ancient artifacts at a site. Each artifact dates back 2000 years.
22
100
true
32
32
128
128
15
An archaeologist discovered five ancient artifacts at a site. Each artifact dates back 2000 years.
22
140
true
32
32
192
192
15
An archaeologist discovered five ancient artifacts at a site. Each artifact dates back 2000 years.
22
184
true
32
32
256
256
15
An archaeologist discovered five ancient artifacts at a site. Each artifact dates back 2000 years.
22
285
true
32
32
384
384
15
An archaeologist discovered five ancient artifacts at a site. Each artifact dates back 2000 years.
22
380
true
32
32
512
512
15
An archaeologist discovered five ancient artifacts at a site. Each artifact dates back 2000 years.
22
580
true
32
32
768
768
15
An archaeologist discovered five ancient artifacts at a site. Each artifact dates back 2000 years.
22
685
true
32
32
1,024
1,024
15
An archaeologist discovered five ancient artifacts at a site. Each artifact dates back 2000 years.
22
1,157
true
32
32
1,536
1,536
15
An archaeologist discovered five ancient artifacts at a site. Each artifact dates back 2000 years.
22
1,716
true
32
32
2,048
2,048
15
A baker has twelve cupcakes. He sells three cupcakes. How many cupcakes does he have left?
22
21
true
32
32
32
32
16
A baker has twelve cupcakes. He sells three cupcakes. How many cupcakes does he have left?
22
30
true
32
32
48
48
16
A baker has twelve cupcakes. He sells three cupcakes. How many cupcakes does he have left?
22
41
true
32
32
64
64
16
A baker has twelve cupcakes. He sells three cupcakes. How many cupcakes does he have left?
22
108
true
32
32
128
128
16
A baker has twelve cupcakes. He sells three cupcakes. How many cupcakes does he have left?
22
144
true
32
32
192
192
16
A baker has twelve cupcakes. He sells three cupcakes. How many cupcakes does he have left?
22
232
true
32
32
256
256
16
A baker has twelve cupcakes. He sells three cupcakes. How many cupcakes does he have left?
22
285
true
32
32
384
384
16
A baker has twelve cupcakes. He sells three cupcakes. How many cupcakes does he have left?
22
392
true
32
32
512
512
16
A baker has twelve cupcakes. He sells three cupcakes. How many cupcakes does he have left?
22
480
true
32
32
768
768
16
A baker has twelve cupcakes. He sells three cupcakes. How many cupcakes does he have left?
22
675
true
32
32
1,024
1,024
16
A baker has twelve cupcakes. He sells three cupcakes. How many cupcakes does he have left?
22
975
true
32
32
1,536
1,536
16
A baker has twelve cupcakes. He sells three cupcakes. How many cupcakes does he have left?
22
1,613
true
32
32
2,048
2,048
16
A person consistently avoids social situations due to anxiety. Identify this behavior as a psychological response to fear.
23
25
true
32
32
32
32
18
A person consistently avoids social situations due to anxiety. Identify this behavior as a psychological response to fear.
23
41
true
32
32
48
48
18
A person consistently avoids social situations due to anxiety. Identify this behavior as a psychological response to fear.
23
51
true
32
32
64
64
18
A person consistently avoids social situations due to anxiety. Identify this behavior as a psychological response to fear.
23
105
true
32
32
128
128
18
A person consistently avoids social situations due to anxiety. Identify this behavior as a psychological response to fear.
23
145
true
32
32
192
192
18
A person consistently avoids social situations due to anxiety. Identify this behavior as a psychological response to fear.
23
170
true
32
32
256
256
18
A person consistently avoids social situations due to anxiety. Identify this behavior as a psychological response to fear.
23
262
true
32
32
384
384
18
A person consistently avoids social situations due to anxiety. Identify this behavior as a psychological response to fear.
23
407
true
32
32
512
512
18
A person consistently avoids social situations due to anxiety. Identify this behavior as a psychological response to fear.
23
550
true
32
32
768
768
18
A person consistently avoids social situations due to anxiety. Identify this behavior as a psychological response to fear.
23
662
true
32
32
1,024
1,024
18
A person consistently avoids social situations due to anxiety. Identify this behavior as a psychological response to fear.
23
1,217
true
32
32
1,536
1,536
18
A person consistently avoids social situations due to anxiety. Identify this behavior as a psychological response to fear.
23
1,550
true
32
32
2,048
2,048
18
In a democratic system, citizens vote for representatives. If 60% of voters support candidate A, what percentage opposes candidate A?
28
22
true
32
32
32
32
20
In a democratic system, citizens vote for representatives. If 60% of voters support candidate A, what percentage opposes candidate A?
28
32
true
32
32
48
48
20
In a democratic system, citizens vote for representatives. If 60% of voters support candidate A, what percentage opposes candidate A?
28
52
true
32
32
64
64
20
In a democratic system, citizens vote for representatives. If 60% of voters support candidate A, what percentage opposes candidate A?
28
88
true
32
32
128
128
20
In a democratic system, citizens vote for representatives. If 60% of voters support candidate A, what percentage opposes candidate A?
28
123
true
32
32
192
192
20
In a democratic system, citizens vote for representatives. If 60% of voters support candidate A, what percentage opposes candidate A?
28
231
true
32
32
256
256
20
In a democratic system, citizens vote for representatives. If 60% of voters support candidate A, what percentage opposes candidate A?
28
257
true
32
32
384
384
20
In a democratic system, citizens vote for representatives. If 60% of voters support candidate A, what percentage opposes candidate A?
28
345
true
32
32
512
512
20
In a democratic system, citizens vote for representatives. If 60% of voters support candidate A, what percentage opposes candidate A?
28
645
true
32
32
768
768
20
In a democratic system, citizens vote for representatives. If 60% of voters support candidate A, what percentage opposes candidate A?
28
913
true
32
32
1,024
1,024
20
In a democratic system, citizens vote for representatives. If 60% of voters support candidate A, what percentage opposes candidate A?
28
1,158
true
32
32
1,536
1,536
20
In a democratic system, citizens vote for representatives. If 60% of voters support candidate A, what percentage opposes candidate A?
28
1,457
true
32
32
2,048
2,048
20
A pea plant with yellow seeds (Y) is crossed with a plant having green seeds (y). What is the genotype?
24
20
true
32
32
32
32
20
A pea plant with yellow seeds (Y) is crossed with a plant having green seeds (y). What is the genotype?
24
40
true
32
32
48
48
20
A pea plant with yellow seeds (Y) is crossed with a plant having green seeds (y). What is the genotype?
24
47
true
32
32
64
64
20
A pea plant with yellow seeds (Y) is crossed with a plant having green seeds (y). What is the genotype?
24
95
true
32
32
128
128
20
End of preview. Expand in Data Studio

Metrum AI Prompt Library

A prompt library for LLM inference workload and performance benchmarking, prepared for use with metrum-ai/bench-cli. It contains 593,730 records with prompt text, intended lengths, token buckets, and reasoning labels. It contains no reference answers.

The full configuration preserves all source records, including repeated prompts and their distinct workload targets. Prompts may appear duplicated, with only target_output_length differing. These variants are intentional: the consuming application should append the intended output word count as a hint to the LLM, for example:

request_prompt = (
    row["prompt"]
    + f"\n\nPlease aim for approximately {row['target_output_length']} words in your response."
)

Thus the same base prompt can exercise different requested output lengths. The hint guides generation; it does not guarantee an exact response length. Keep the stored prompt unchanged and append the hint when constructing the request. Include the appended hint when measuring the rendered input length.

The sample provides 2,960 records, ten from each of 296 observed input-bucket/output-bucket/reasoning groups. It is a balanced smoke-test sample, not a frequency-representative sample of the full library. The train split name is a loading convention, not a recommendation to train on this benchmark library.

Load 

from datasets import load_dataset

# For reproducible benchmarking, replace main with a commit SHA from the Files tab.
revision = "main"
ds = load_dataset("metrum-ai/prompt-library", "full",
                  split="train", revision=revision, streaming=True)
print(next(iter(ds)))

sample = load_dataset("metrum-ai/prompt-library", "sample",
                      split="train", revision=revision)

The full configuration uses 12 Zstandard-compressed Parquet shards with at most 50,000 rows per shard. Both configurations use explicit file mappings, so loading them does not also load the raw JSONL or prompt-only sample export. See Hugging Face loading documentation.

Fields

Field Type Meaning and limitations
prompt string Original prompt text, preserved exactly.
prompt_length int64 Source-supplied legacy length; counting method is unverified.
target_output_length int64 Intended output word count, to append to the prompt as a generation hint. Confirmed by the dataset owner; not measured output.
reasoning bool Source reasoning label. Strings "true" and "false" normalized to booleans. Not a model setting or validated difficulty score.
target_input_tokens int64 Source-supplied input token target; originating tokenizer unknown.
input_token_bucket int64 Source input-token bucket.
target_output_tokens int64 Source-supplied output token target; not a generated length.
output_token_bucket int64 Source output-token bucket.
actual_words int64 Source-supplied input word-count metadata; counting method is unverified.

All numeric metadata is retained unchanged. prompt_length and actual_words differ in 568,784 records. For actual input words, compute len(prompt.split()) using Unicode whitespace splitting. For model-specific token lengths, count using an identified tokenizer and state whether system/chat formatting is included. Do not infer a universal tokens-per-word ratio.

Token targets currently equal their corresponding buckets in every record, but that does not establish that they equal actual token counts. Target input tokens range from 32 to 7,000, and target output tokens from 32 to 2,048. See manifest.json for all group counts and numeric statistics.

Use with bench-cli

Get the CLI, installation instructions, and usage documentation from metrum-ai/bench-cli on GitHub.

The inspected CLI accepts local or HTTP(S) JSONL with a string prompt field, extracts only that field, and uses a global --max-tokens value. Consequently, the raw source and the prompt-only sample.jsonl already satisfy its current loader contract. The CLI does not yet use this dataset's per-row length targets or append the word-count hint. A consuming application or the future bench-cli integration must construct that hint from each full dataset row; the prompt-only sample export does not contain the metadata. Its existing loader also reads the full input into memory; start with the small sample.

Example for the existing CLI, using an endpoint/model you supply:

metrum-ai-bench-llm \
  --scenario prompt-library-smoke \
  --url "$ENDPOINT_URL" --model "$MODEL" \
  --prompts sample.jsonl --mode chat \
  --num-requests 100 --concurrency 1 --max-tokens 128 \
  --seed 42 --data-log results.jsonl

This example uses a uniform token cap; it does not target a particular mix. Download sample.jsonl from the dataset's Files tab first, and configure any endpoint authentication as required by your bench-cli version.

BENCH_CLI_HANDOFF.md is an implementation brief for an agent in the bench-cli repository. It specifies pinned dataset loading, joint ISL/OSL mean or median selection, words or tokens as explicit units, per-request budgets, deterministic schedules, and requested-versus-observed statistics. These features require subsequent changes to bench-cli.

Source preservation and reproducibility

  • raw/prompt-library-new.jsonl is the exact original 2,028,539,356-byte JSONL file, retained without changes.
  • Parquet preserves all nine fields and source row order; the only value transformation is normalization of 15,872 string reasoning labels to booleans.
  • Distinct rows are retained even when their prompt text matches. There are 526,912 distinct prompt strings, identified during preparation by SHA-256.
  • The sample ranks each source row by SHA256("42:<one-based-source-line>") within its normalized group, takes the ten lowest ranks, and restores source order. No prompt text or numeric metadata is rewritten.
  • sample-index.json maps each sample row to its one-based source line. Full-row zero-based ordinals are source lines minus one.
  • manifest.json records source checksum, counts, and group distributions. checksums.sha256 covers the uploaded package files except itself and Hub-generated metadata.

See VALIDATION.md for the preparation and compatibility checks.

Rebuild into a new output directory using Python 3.12:

python -m pip install -r requirements.txt
python scripts/prepare_dataset.py raw/prompt-library-new.jsonl ../rebuilt
python scripts/validate_dataset.py ../rebuilt
python -m unittest discover -s scripts -p 'test_*.py'

The preparation script streams the source and writes bounded-size batches. The validator compares every source record with the Parquet output and independently reconstructs the sample selection. Use scripts/publish_dataset.py to upload an explicitly allowlisted package; it never uploads credential files.

Provenance, intended use, and license

The source file was supplied by the dataset publisher. Its original authors, generation process, source collection, and tokenizer are not documented in the provided material. The owner confirmed the meaning of target_output_length and selected Apache-2.0 for publication; see LICENSE.

This is a performance-workload library, not an answer-quality benchmark. Reasoning labels and requested lengths have not been independently validated. There has been no content-safety, privacy, representativeness, or factual-quality audit of the prompt corpus. Generated responses may stop well before token limits, and actual output lengths must be measured separately from targets.
Downloads last month
696
Number of rows:
596,690
Total file size:
2.17 GB