Vision language models are blind
Paper • 2407.06581 • Published • 84
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.