| { |
| "QformConfig": { |
| "encoder_hidden_size": 768, |
| "intermediate_size": 2048, |
| "model_type": "blip_2_qformer", |
| "num_attention_heads": 8, |
| "num_hidden_layers": 8 |
| }, |
| "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 |
| }, |
| "architectures": [ |
| "QformT5ForVQA" |
| ], |
| "continue_training": true, |
| "from_pretrained_qform": null, |
| "from_pretrained_t5": "VietAI/vit5-base", |
| "model_type": "qformt5", |
| "patch_per_side": 24, |
| "torch_dtype": "float32", |
| "transformers_version": "4.36.0", |
| "visual_dmodel": 768 |
| } |
|
|
