model.json

{
  "format": "onnx",
  "model_type": "vlm",
  "architecture": "fastvlm-qwen2",
  "encoder": {
    "file": "onnx/vision_encoder.onnx",
    "type": "vision_encoder",
    "inputs": {
      "pixel_values": {
        "dtype": "float32",
        "shape": ["batch", 3, "height", "width"]
      }
    },
    "outputs": {
      "image_features": {
        "dtype": "float32",
        "shape": ["batch", "tokens", "hidden_dim"]
      }
    }
  },
  "decoder": {
    "file": "onnx/decoder_model_merged.onnx",
    "type": "causal_lm",
    "inputs": {
      "input_ids": {
        "dtype": "int64",
        "shape": ["batch", "sequence"]
      },
      "attention_mask": {
        "dtype": "int64",
        "shape": ["batch", "sequence"]
      }
    },
    "outputs": {
      "logits": {
        "dtype": "float32",
        "shape": ["batch", "sequence", "vocab_size"]
      }
    }
  },
  "embed_tokens": {
    "file": "onnx/embed_tokens.onnx",
    "type": "embedding",
    "inputs": {
      "input_ids": {
        "dtype": "int64",
        "shape": ["batch", "sequence"]
      }
    },
    "outputs": {
      "embeddings": {
        "dtype": "float32",
        "shape": ["batch", "sequence", "hidden_dim"]
      }
    }
  },
  "tokenizer": {
    "file": "tokenizer.json",
    "vocab_file": "vocab.json",
    "merges_file": "merges.txt",
    "special_tokens_map": "special_tokens_map.json",
    "tokenizer_config": "tokenizer_config.json"
  },
  "generation_config": "generation_config.json",
  "preprocessor": {
    "config": "preprocessor_config.json",
    "processor_config": "processor_config.json"
  }
}