Where is Image to text projector

#3
by rv2307 - opened

As you are following Llava's approach you need a mm projector to conver the image space to text space. But how are you doing this without projector, Or is projector available in the model bin?

They are in the model bin / safe tensor files:

"multi_modal_projector.linear_1.bias": "model-00001-of-00003.safetensors",
"multi_modal_projector.linear_1.weight": "model-00001-of-00003.safetensors",
"multi_modal_projector.linear_2.bias": "model-00001-of-00003.safetensors",
"multi_modal_projector.linear_2.weight": "model-00001-of-00003.safetensors",

matthewlyleolson changed discussion status to closed

Sign up or log in to comment