cmu/refav
Multimodal
Coding
Autonomous-vehicle scenario mining via VLM.
Run this task
CLI:
inspect eval inspect_harbor/cmu_refav --model openai/gpt-5Python:
from inspect_ai import eval
from inspect_harbor import cmu_refav
eval(cmu_refav(), model="openai/gpt-5")Dataset information
| Harbor registry | cmu/refav |
| Inspect task | cmu_refav |
| Latest digest | sha256:57c93f1c2f6bda2b015cec8286380cbd33860c6d1d9407f1d17dd656c0b926e0 |
| Samples | 1000 |
| Paper | arxiv |
| Source | https://github.com/CainanD/RefAV |
See Task Parameters for the parameter set shared across all Harbor tasks.