CSV preview
The table below displays the original header and example row without changing the downloadable data.
| candidate | model_identifier | runtime_version | quantization | task | prompt_hash | context | temperature | output_limit | expected_result | observed_result | stop_reason | latency_ms | memory_peak | decision | test_date |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| example | qwen2.5:1.5b | Ollama 0.35.1 | not recorded | source-verification bullets | not recorded | 2048 | 0 | 80 | five complete bullets | CPU produced five bullets | stop | 26175.9 | NOT_TESTED | tested only for this bounded prompt | 2026-10-27 |
Column guide
candidate- A short name for the evaluated candidate.
model_identifier- The exact model name or identifier.
runtime_version- The runtime and version used for the test.
quantization- The model quantization, when recorded.
task- The bounded task used for comparison.
prompt_hash- A repeatability identifier for the prompt.
context- The configured context length.
temperature- The sampling temperature.
output_limit- The maximum generated output.
expected_result- The pass condition defined before the run.
observed_result- What the run actually produced.
stop_reason- Why generation stopped.
latency_ms- Measured elapsed time in milliseconds.
memory_peak- Observed peak memory use, or NOT_TESTED.
decision- The conclusion supported by this test.
test_date- The date of the recorded test.
Use in Excel or LibreOffice
- Select Download to save the original CSV file.
- Open the file in Excel or LibreOffice Calc and confirm comma-separated import if prompted.
- Keep the header row and replace the synthetic example row with your own reviewed entries.
- Save a working copy so the original template remains available.
candidate,model_identifier,runtime_version,quantization,task,prompt_hash,context,temperature,output_limit,expected_result,observed_result,stop_reason,latency_ms,memory_peak,decision,test_date
example,qwen2.5:1.5b,Ollama 0.35.1,not recorded,source-verification bullets,not recorded,2048,0,80,five complete bullets,CPU produced five bullets,stop,26175.9,NOT_TESTED,tested only for this bounded prompt,2026-10-27