Test fixture
Arithmetic, symbolic steps, and structured problem solving.
The model receives the prompt (and optional system message). The run uses scorer rubric_json_metrics with the JSON configuration below. Pass/fail and partial credit are determined entirely by that scorer against the model output; no human grading.
Return JSON only with keys answer, evidence, constraints. Error rate spikes after renaming retryDelayMs to retry_delay_ms. Logs show delay undefined and retries immediate. Identify root cause.
{
"metrics": {
"accuracy": {
"checks": [
{
"contains": [
"retryDelayMs"
]
},
{
"contains": [
"retry_delay_ms"
]
},
{
"contains": [
"undefined"
]
}
]
},
"evidence": {
"checks": [
{
"contains": [
"immediate retries"
]
},
{
"contains": [
"error rate"
]
},
{
"contains": [
"config"
]
}
]
},
"constraint": {
"checks": [
{
"contains": [
"root cause"
]
},
{
"contains": [
"rename"
]
},
{
"contains": [
"retry storm"
]
}
]
}
}
}temperature
0
max_tokens
500
timeout (s)
120
type
scored
file
reason-root-retry-storm.json