Vision language models are blind
Paper
•
2407.06581
•
Published
•
83
VLM Performance
Note Use the BlindTest Eval benchmark for vision tasks that are easy for humans.
Note Use the BlindTest Eval benchmark for vision tasks that are easy for humans.