bliva/configs/models/bliva_flant5xxl.yaml

model:
  arch: flant5xxl
  load_finetuned: True
  load_pretrained: False

  pretrained: "https://storage.googleapis.com/sfr-vision-language-research/LAVIS/models/InstructBLIP/instruct_blip_flanxxl_trimmed.pth"
  finetuned: ''

  # vit encoder
  image_size: 224
  drop_path_rate: 0
  use_grad_checkpoint: False
  vit_precision: "fp16"
  freeze_vit: True

  # Q-Former
  num_query_token: 32

  # T5
  t5_model: "google/flan-t5-xxl"

  # generation configs
  prompt: ""


preprocess:
    vis_processor:
        train:
          name: "blip_image_train"
          image_size: 224
        eval:
          name: "blip_image_eval"
          image_size: 224
    text_processor:
        train:
          name: "blip_caption"
        eval:
          name: "blip_caption"