smpanaro's picture
Update sequoia mode with transposed value cache and 4:508 input:cache length
722eedf verified
[
{
"metadataOutputVersion" : "3.0",
"outputSchema" : [
{
"hasShapeFlexibility" : "0",
"isOptional" : "0",
"dataType" : "Int32",
"formattedType" : "MultiArray (Int32)",
"shortDescription" : "",
"shape" : "[]",
"name" : "argmax",
"type" : "MultiArray"
}
],
"modelParameters" : [
],
"specificationVersion" : 7,
"mlProgramOperationTypeHistogram" : {
"Ios16.reduceArgmax" : 1
},
"computePrecision" : "Mixed (Float16, Int32)",
"isUpdatable" : "0",
"stateSchema" : [
],
"availability" : {
"macOS" : "13.0",
"tvOS" : "16.0",
"visionOS" : "1.0",
"watchOS" : "9.0",
"iOS" : "16.0",
"macCatalyst" : "16.0"
},
"modelType" : {
"name" : "MLModelType_mlProgram"
},
"userDefinedMetadata" : {
"com.github.apple.coremltools.source_dialect" : "TorchScript",
"com.github.apple.coremltools.source" : "torch==2.1.0",
"com.github.apple.coremltools.version" : "8.0b1"
},
"inputSchema" : [
{
"shortDescription" : "",
"dataType" : "Float16",
"hasShapeFlexibility" : "1",
"isOptional" : "0",
"shapeFlexibility" : "1 × 511 × 32000 | 1 × 1 × 32000 | 1 × 2 × 32000 | 1 × 4 × 32000 | 1 × 64 × 32000 | 1 × 512 × 32000",
"formattedType" : "MultiArray (Float16 1 × 511 × 32000)",
"type" : "MultiArray",
"shape" : "[1, 511, 32000]",
"name" : "logits",
"enumeratedShapes" : "[[1, 511, 32000], [1, 1, 32000], [1, 2, 32000], [1, 4, 32000], [1, 64, 32000], [1, 512, 32000]]"
}
],
"generatedClassName" : "logit_processor_2",
"method" : "predict"
}
]