OnnxSpec

class OnnxSpec(val visualFile: String, val textualFile: String, val size: Int, val mean: FloatArray, val std: FloatArray, val centerCrop: Boolean, val dim: Int, val tokenizerFiles: List<String>, val tokenizer: (File) -> (String) -> IntArray)(source)

How an ONNX image+text model wants its inputs.

Constructors

Link copied to clipboard
constructor(visualFile: String, textualFile: String, size: Int, mean: FloatArray, std: FloatArray, centerCrop: Boolean, dim: Int, tokenizerFiles: List<String>, tokenizer: (File) -> (String) -> IntArray)

Properties

Link copied to clipboard

Shortest side to size then center crop; false squashes the whole photo to a square.

Link copied to clipboard
val dim: Int
Link copied to clipboard
Link copied to clipboard
Link copied to clipboard
val size: Int
Link copied to clipboard
Link copied to clipboard
Link copied to clipboard
val tokenizer: (File) -> (String) -> IntArray
Link copied to clipboard
Link copied to clipboard