Skip to content
Open
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 4 additions & 1 deletion ifstruct/eval.py
Original file line number Diff line number Diff line change
Expand Up @@ -34,14 +34,15 @@ class EvalResult:
require_wrapper_key: bool


def run_one(example: IfStructExample, *, model: str, base_url: str, api_key: str, max_tokens: int, temperature: float, max_retries: int = 40) -> EvalResult:
def run_one(example: IfStructExample, *, model: str, base_url: str, api_key: str, max_tokens: int, temperature: float, timeout: float, max_retries: int = 40) -> EvalResult:
completion = chat_completion(
base_url=base_url,
api_key=api_key,
model=model,
prompt=example.prompt,
max_tokens=max_tokens,
temperature=temperature,
timeout=timeout,
max_retries=max_retries,
)
validation = validate_response(
Expand Down Expand Up @@ -243,6 +244,7 @@ def main() -> None:
parser.add_argument("--offset", type=int, default=0, help="Optional offset into the dataset.")
parser.add_argument("--max-tokens", type=int, default=16000)
parser.add_argument("--temperature", type=float, default=0.0)
parser.add_argument("--timeout", type=float, default=120.0, help="Request timeout in seconds.")
parser.add_argument("--max-retries", type=int, default=40, help="Max API retries per sample.")
parser.add_argument("--seed", type=int, nargs="+", default=None, help="Run only these seed(s).")
parser.add_argument("--verbose", "-v", action="store_true")
Expand Down Expand Up @@ -283,6 +285,7 @@ def main() -> None:
api_key=args.api_key,
max_tokens=args.max_tokens,
temperature=args.temperature,
timeout=args.timeout,
max_retries=args.max_retries,
): example
for example in examples
Expand Down