53 lines
1.9 KiB
JSON
53 lines
1.9 KiB
JSON
{
|
|
"traces": [],
|
|
"train": [],
|
|
"demos": [],
|
|
"signature": {
|
|
"instructions": "Evaluate whether this writing is good or bad for the life simulation game.\n\nJudge it as a deeply human problem: whether it is good enough, specific enough, and rewarding enough for the player.\n\nIdentify when the writing is too wordy, too cute with its phrasing, or dangles off into something nonspecific.\n\nPrefer writing that feels specific, meaningful, and reaches a logical arc conclusion rather than stopping at a flat or unrewarding next thing that happens.\n\nReturn a qualitative score, whether it passes overall, the issues or none if it passes.",
|
|
"fields": [
|
|
{
|
|
"prefix": "Text:",
|
|
"description": "The generated writing to evaluate."
|
|
},
|
|
{
|
|
"prefix": "Scenario Context:",
|
|
"description": "Context about the event, choice, or outcome so the writing can be judged in context."
|
|
},
|
|
{
|
|
"prefix": "Writing Quality Score:",
|
|
"description": "A qualitative score for the writing (0-10)."
|
|
},
|
|
{
|
|
"prefix": "Passes:",
|
|
"description": "Whether the writing passes overall."
|
|
},
|
|
{
|
|
"prefix": "Reasoning:",
|
|
"description": "Your reasoning for your answer. Inlude any uncertainties about your answer or ambiguity in the task."
|
|
},
|
|
{
|
|
"prefix": "Issues:",
|
|
"description": "Problems identified in the writing."
|
|
}
|
|
]
|
|
},
|
|
"lm": {
|
|
"model": "openrouter/openai/gpt-oss-120b",
|
|
"model_type": "chat",
|
|
"cache": true,
|
|
"num_retries": 3,
|
|
"finetuning_model": null,
|
|
"launch_kwargs": {},
|
|
"train_kwargs": {},
|
|
"temperature": null,
|
|
"max_tokens": null,
|
|
"base_url": "https://openrouter.ai/api/v1"
|
|
},
|
|
"metadata": {
|
|
"dependency_versions": {
|
|
"python": "3.13",
|
|
"dspy": "3.2.0",
|
|
"cloudpickle": "3.1"
|
|
}
|
|
}
|
|
} |