metadata
license: apache-2.0
This model was trained for 5 epochs on 3 different tasks in a curriculum learning fashion. The first task was object classification for text-object level alignment, followed by referring region description and finally object instruction fllowing. The LLM decoder backbone is llama-2-7b-hf and the vision encoder is a clip-vit-large-patch14-336 model.
For more details on training and usage, check out the github repository at https://github.com/tossowski/Olive.