implementation detail · filed under evaluation
JSON-structured answer-format prompt
A prompt structure is used to standardize answer formatting for multiple-choice evaluation.
- source
- 1
- model
- 1
- lab adopt it
- 1
- strongest
- used
How sources treat it
One count per evidence span, weakest treatment to strongest.
used 1
Documented in
Evidence
1 span quoted from the sources, strongest treatment first.
Please show your choice in the answerfield with only the choice letter
usedevaluation onlyin Qwen3.5-122B-A10BQwen
Filed alongside
Other methods under evaluation :: evaluation harness.
DeepSeek Harness Minimal modeHarborNeMo Evaluator SDKAtomic binary rubric evaluationAutomatic in-flight checkpoint evaluation schedulingBenchmark evaluation frameworkCross-scaffold evaluationInline training evaluatorsIsolated container per rolloutmini-swe-agent harnessNeMo Gym and NeMo Evaluator-based harnessNeMo SkillsPool