EvalHarness/evalharness/config/dp4-nothink/imo_answerbench.yaml
2026-09-11 07:59:42 +00:00

13 lines
172 B
YAML

generation:
temperature: 1.0
max_tokens: 8192
top_p: 1.0
run:
limit_per_task: 25
concurrency: 8
judge:
model: dp4-flash
api_url: http://174.1.51.4:30000/v1