config.json

{
  "architectures": [
    "BlipForQuestionAnswering"
  ],
  "bos_token_id": 0,
  "eos_token_id": 2,
  "image_text_hidden_size": 256,
  "initializer_factor": 1.0,
  "initializer_range": 0.02,
  "label_smoothing": 0.0,
  "logit_scale_init_value": 2.6592,
  "model_type": "blip",
  "pad_token_id": 0,
  "projection_dim": 64,
  "sep_token_id": 3,
  "text_config": {
    "attention_dropout": 0.1,
    "bos_token_id": 0,
    "dropout": 0.1,
    "encoder_hidden_size": 32,
    "hidden_size": 32,
    "intermediate_size": 37,
    "model_type": "blip_text_model",
    "num_attention_heads": 4,
    "num_hidden_layers": 2,
    "projection_dim": 32,
    "sep_token_id": 3,
    "vocab_size": 1124
  },
  "torch_dtype": "float32",
  "transformers_version": "4.39.0.dev0",
  "vision_config": {
    "attention_dropout": 0.1,
    "dropout": 0.1,
    "hidden_size": 32,
    "image_size": 30,
    "intermediate_size": 37,
    "model_type": "blip_vision_model",
    "num_attention_heads": 4,
    "num_channels": 3,
    "num_hidden_layers": 2,
    "patch_size": 2,
    "projection_dim": 32
  }
}