original/compiled/Unet.mlmodelc/metadata.json

[
  {
    "shortDescription" : "This is a model that can be used to generate and modify images based on text prompts.It is a Latent Diffusion Model that uses two fixed, pretrained text encoders (OpenCLIP-ViT\/G and CLIP-ViT\/L).Please refer to https:\/\/arxiv.org\/abs\/2307.01952 for details",
    "metadataOutputVersion" : "3.0",
    "outputSchema" : [
      {
        "hasShapeFlexibility" : "0",
        "isOptional" : "0",
        "dataType" : "Float32",
        "formattedType" : "MultiArray (Float32)",
        "shortDescription" : "Same shape and dtype as the `sample` input. The predicted noise to facilitate the reverse diffusion (denoising) process",
        "shape" : "[]",
        "name" : "noise_pred",
        "type" : "MultiArray"
      }
    ],
    "version" : "stabilityai\/stable-diffusion-xl-base-0.9",
    "modelParameters" : [

    ],
    "author" : "Please refer to the Model Card available at huggingface.co\/stabilityai\/stable-diffusion-xl-base-0.9",
    "specificationVersion" : 7,
    "storagePrecision" : "Float16",
    "license" : "Please refer to the Model Card available at huggingface.co\/stabilityai\/stable-diffusion-xl-base-0.9\/blob\/main\/LICENSE.md",
    "mlProgramOperationTypeHistogram" : {
      "UpsampleNearestNeighbor" : 2,
      "Ios16.reduceMean" : 512,
      "Ios16.sin" : 2,
      "Ios16.softmax" : 140,
      "Split" : 70,
      "Ios16.add" : 722,
      "Concat" : 14,
      "Ios16.realDiv" : 46,
      "Ios16.square" : 46,
      "ExpandDims" : 6,
      "Ios16.sub" : 256,
      "Ios16.cast" : 1,
      "Ios16.conv" : 794,
      "Ios16.constexprLutToDense" : 870,
      "Ios16.gelu" : 70,
      "Ios16.matmul" : 280,
      "Ios16.batchNorm" : 46,
      "Ios16.reshape" : 676,
      "Ios16.rsqrt" : 210,
      "Ios16.silu" : 38,
      "Ios16.sqrt" : 46,
      "Ios16.mul" : 842,
      "Ios16.cos" : 2,
      "SliceByIndex" : 4
    },
    "computePrecision" : "Mixed (Float32, Float16, Int32)",
    "isUpdatable" : "0",
    "availability" : {
      "macOS" : "13.0",
      "tvOS" : "16.0",
      "watchOS" : "9.0",
      "iOS" : "16.0",
      "macCatalyst" : "16.0"
    },
    "modelType" : {
      "name" : "MLModelType_mlProgram"
    },
    "inputSchema" : [
      {
        "hasShapeFlexibility" : "0",
        "isOptional" : "0",
        "dataType" : "Float16",
        "formattedType" : "MultiArray (Float16 2 × 4 × 128 × 128)",
        "shortDescription" : "The low resolution latent feature maps being denoised through reverse diffusion",
        "shape" : "[2, 4, 128, 128]",
        "name" : "sample",
        "type" : "MultiArray"
      },
      {
        "hasShapeFlexibility" : "0",
        "isOptional" : "0",
        "dataType" : "Float16",
        "formattedType" : "MultiArray (Float16 2)",
        "shortDescription" : "A value emitted by the associated scheduler object to condition the model on a given noise schedule",
        "shape" : "[2]",
        "name" : "timestep",
        "type" : "MultiArray"
      },
      {
        "hasShapeFlexibility" : "0",
        "isOptional" : "0",
        "dataType" : "Float16",
        "formattedType" : "MultiArray (Float16 2 × 2048 × 1 × 77)",
        "shortDescription" : "Output embeddings from the associated text_encoder model to condition to generated image on text. A maximum of 77 tokens (~40 words) are allowed. Longer text is truncated. Shorter text does not reduce computation.",
        "shape" : "[2, 2048, 1, 77]",
        "name" : "encoder_hidden_states",
        "type" : "MultiArray"
      },
      {
        "hasShapeFlexibility" : "0",
        "isOptional" : "0",
        "dataType" : "Float16",
        "formattedType" : "MultiArray (Float16 2 × 1280)",
        "shortDescription" : "Additional embeddings passed to the unet based on the pooled output of the text encoders.",
        "shape" : "[2, 1280]",
        "name" : "text_embeds",
        "type" : "MultiArray"
      },
      {
        "hasShapeFlexibility" : "0",
        "isOptional" : "0",
        "dataType" : "Float16",
        "formattedType" : "MultiArray (Float16 2 × 6)",
        "shortDescription" : "Additional embeddings passed to the unet based on width and height dimensions.For SDXL, default values look like [1024, 1024, 0, 0, 1024, 1024]",
        "shape" : "[2, 6]",
        "name" : "time_ids",
        "type" : "MultiArray"
      }
    ],
    "userDefinedMetadata" : {
      "com.github.apple.coremltools.version" : "7.0b1",
      "com.github.apple.coremltools.source" : "torch==2.1.0.dev20230718",
      "com.github.apple.ml-stable-diffusion.version" : "1.0.0"
    },
    "generatedClassName" : "Stable_Diffusion_version_stabilityai_stable_diffusion_xl_base_0_9_unet",
    "method" : "predict"
  }
]