{
  "T5ModelConfig": {
    "architectures": [
      "T5ForConditionalGeneration"
    ],
    "d_ff": 3072,
    "d_model": 768,
    "decoder_start_token_id": 0,
    "model_type": "t5",
    "n_positions": 512,
    "num_decoder_layers": 12,
    "num_heads": 12,
    "num_layers": 12,
    "output_past": true,
    "torch_dtype": "float32",
    "vocab_size": 36096
  },
  "TextQformConfig": {
    "encoder_hidden_size": 768,
    "hidden_size": 768,
    "intermediate_size": 2048,
    "model_type": "blip_2_qformer",
    "num_hidden_layers": 6
  },
  "VisualPerceiverConfig": {
    "d_latents": 768,
    "model_type": "perceiver",
    "num_latents": 100,
    "num_self_attends_per_block": 16
  },
  "VisualQformConfig": {
    "encoder_hidden_size": 768,
    "hidden_size": 768,
    "intermediate_size": 2048,
    "model_type": "blip_2_qformer",
    "num_hidden_layers": 6
  },
  "architectures": [
    "SepeT5ForVQA"
  ],
  "continue_training": true,
  "from_pretrained_t5": "VietAI/vit5-base",
  "model_type": "qformt5",
  "torch_dtype": "float32",
  "transformers_version": "4.36.0",
  "visual_dmodel": 768
}