Test fixture
Arithmetic, symbolic steps, and structured problem solving.
The model receives the prompt (and optional system message). The run uses scorer json_field_equals with the JSON configuration below. Pass/fail and partial credit are determined entirely by that scorer against the model output; no human grading.
Return valid JSON only with key result. Compute: ((83 + 51) * 27) - 25.
{
"field": "result",
"value": 3593
}temperature
0
max_tokens
60
timeout (s)
120
type
scored
file
reasoning_hard_20.json