bigcodebench-hard-complete@1.0.0

Coding

BigCodeBench-Hard complete benchmark adapter for Harbor - challenging Python programming tasks with reward-based verification.

← Back to Registry

Run this task

CLI:

inspect eval inspect_harbor/bigcodebench_hard_complete_1_0_0 --model openai/gpt-5

Python:

from inspect_ai import eval
from inspect_harbor import bigcodebench_hard_complete_1_0_0

eval(bigcodebench_hard_complete_1_0_0(), model="openai/gpt-5")

Dataset information

Harbor registry bigcodebench-hard-complete@1.0.0
Inspect task bigcodebench_hard_complete_1_0_0
Version 1.0.0
Samples 145
Paper arxiv

See Task Parameters for the parameter set shared across all Harbor tasks.