Hugging Face's logo

Datasets:
RKB109
/
production-ai-observability-20260731-dataset

Tasks:
Text Classification
Modalities:
Text
Formats:
json
Languages:
English
Size:
< 1K
Tags:
Synthetic
ai-observability
evaluation
text-classification
token-classification
summarization
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
1
Dataset Viewer
Auto-converted to Parquet Duplicate
id
stringlengths
20
20
input
stringlengths
46
81
label
stringclasses
4 values
context
stringclasses
6 values
source
stringclasses
6 values
variant
stringclasses
3 values
synthetic
bool
1 class
ai-observability-1-1
Request latency rose above the service objective
latency-regression
End-to-end latency exceeded the approved percentile threshold.
trace-01
direct
true
ai-observability-1-2
In an operations review, Request latency rose above the service objective
latency-regression
End-to-end latency exceeded the approved percentile threshold.
trace-01
operations
true
ai-observability-1-3
For an evaluation case, Request latency rose above the service objective
latency-regression
End-to-end latency exceeded the approved percentile threshold.
trace-01
evaluation
true
ai-observability-2-2
In an operations review, Prompt tokens doubled after a template change
token-spike
Token consumption increased beyond the cost and context baseline.
trace-02
operations
true
ai-observability-2-3
For an evaluation case, Prompt tokens doubled after a template change
token-spike
Token consumption increased beyond the cost and context baseline.
trace-02
evaluation
true
ai-observability-3-1
The retrieval tool returned a timeout exception
tool-failure
A required external tool failed during execution.
trace-03
direct
true
ai-observability-3-3
For an evaluation case, The retrieval tool returned a timeout exception
tool-failure
A required external tool failed during execution.
trace-03
evaluation
true
ai-observability-4-1
Grounded answer score dropped after deployment
quality-regression
Evaluation quality regressed relative to the release baseline.
trace-04
direct
true
ai-observability-4-2
In an operations review, Grounded answer score dropped after deployment
quality-regression
Evaluation quality regressed relative to the release baseline.
trace-04
operations
true
ai-observability-5-1
The agent stayed within quality limits but became slower
latency-regression
Performance changed without a matching quality improvement.
trace-05
direct
true
ai-observability-5-2
In an operations review, The agent stayed within quality limits but became slower
latency-regression
Performance changed without a matching quality improvement.
trace-05
operations
true
ai-observability-5-3
For an evaluation case, The agent stayed within quality limits but became slower
latency-regression
Performance changed without a matching quality improvement.
trace-05
evaluation
true
ai-observability-6-2
In an operations review, Search calls failed with repeated connection errors
tool-failure
Dependency errors prevented the workflow from completing.
trace-06
operations
true
ai-observability-6-3
For an evaluation case, Search calls failed with repeated connection errors
tool-failure
Dependency errors prevented the workflow from completing.
trace-06
evaluation
true

Production AI Observability Monitor Synthetic Dataset

Summary

This dataset contains 14 training examples and 4 held-out examples for Production AI teams need trace-level signals for latency, token growth, tool failures, and low-quality outputs.

Every record is synthetic and includes:

  • input: query, event, or feature description
  • label: expected class, route, relation, or evidence category
  • context: synthetic supporting context
  • source: fictional source identifier
  • variant: generation pattern
  • synthetic: always true

Uses

  • Reproducible unit and integration tests
  • Baseline model training
  • Evaluation harness development
  • Schema and architecture demonstrations

Limitations

Thresholds are demonstration defaults and need calibration against each production workload.

This dataset does not represent real users, patients, customers, production traffic, or licensed media. It must not be presented as real-world evidence.

Related Model

RKB109/production-ai-observability-20260731-model
Downloads last month
32
Number of rows:
18
Total file size:
11.1 kB

Models trained or fine-tuned on RKB109/production-ai-observability-20260731-dataset