arda-argmax's picture
Upload model distil-whisper_distil-large-v3_turbo_600MB - QLoRA compressed encoder variant
95db02e verified
raw
history blame
4.41 kB
[
{
"metadataOutputVersion" : "3.0",
"storagePrecision" : "Float16",
"outputSchema" : [
{
"hasShapeFlexibility" : "0",
"isOptional" : "0",
"dataType" : "Float16",
"formattedType" : "MultiArray (Float16 1 × 1 × 51866)",
"shortDescription" : "",
"shape" : "[1, 1, 51866]",
"name" : "logits",
"type" : "MultiArray"
},
{
"hasShapeFlexibility" : "0",
"isOptional" : "0",
"dataType" : "Float16",
"formattedType" : "MultiArray (Float16 1 × 2560 × 1 × 1)",
"shortDescription" : "",
"shape" : "[1, 2560, 1, 1]",
"name" : "key_cache_updates",
"type" : "MultiArray"
},
{
"hasShapeFlexibility" : "0",
"isOptional" : "0",
"dataType" : "Float16",
"formattedType" : "MultiArray (Float16 1 × 2560 × 1 × 1)",
"shortDescription" : "",
"shape" : "[1, 2560, 1, 1]",
"name" : "value_cache_updates",
"type" : "MultiArray"
}
],
"modelParameters" : [
],
"specificationVersion" : 7,
"mlProgramOperationTypeHistogram" : {
"Split" : 2,
"Concat" : 2,
"Ios16.rsqrt" : 7,
"Ios16.mul" : 26,
"Squeeze" : 1,
"Ios16.sub" : 8,
"Transpose" : 1,
"Ios16.conv" : 20,
"Ios16.add" : 20,
"Ios16.linear" : 1,
"Ios16.matmul" : 8,
"Ios16.gelu" : 2,
"Ios16.reduceMean" : 14,
"ExpandDims" : 6,
"Ios16.batchNorm" : 7,
"Ios16.gather" : 2,
"Ios16.reshape" : 16,
"Ios16.softmax" : 4
},
"computePrecision" : "Mixed (Float16, Int32)",
"isUpdatable" : "0",
"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==2.2.1",
"com.github.apple.coremltools.version" : "7.1"
},
"inputSchema" : [
{
"hasShapeFlexibility" : "0",
"isOptional" : "0",
"dataType" : "Int32",
"formattedType" : "MultiArray (Int32 1)",
"shortDescription" : "",
"shape" : "[1]",
"name" : "input_ids",
"type" : "MultiArray"
},
{
"hasShapeFlexibility" : "0",
"isOptional" : "0",
"dataType" : "Int32",
"formattedType" : "MultiArray (Int32 1)",
"shortDescription" : "",
"shape" : "[1]",
"name" : "cache_length",
"type" : "MultiArray"
},
{
"hasShapeFlexibility" : "0",
"isOptional" : "0",
"dataType" : "Float16",
"formattedType" : "MultiArray (Float16 1 × 2560 × 1 × 224)",
"shortDescription" : "",
"shape" : "[1, 2560, 1, 224]",
"name" : "key_cache",
"type" : "MultiArray"
},
{
"hasShapeFlexibility" : "0",
"isOptional" : "0",
"dataType" : "Float16",
"formattedType" : "MultiArray (Float16 1 × 2560 × 1 × 224)",
"shortDescription" : "",
"shape" : "[1, 2560, 1, 224]",
"name" : "value_cache",
"type" : "MultiArray"
},
{
"hasShapeFlexibility" : "0",
"isOptional" : "0",
"dataType" : "Float16",
"formattedType" : "MultiArray (Float16 1 × 224)",
"shortDescription" : "",
"shape" : "[1, 224]",
"name" : "kv_cache_update_mask",
"type" : "MultiArray"
},
{
"hasShapeFlexibility" : "0",
"isOptional" : "0",
"dataType" : "Float16",
"formattedType" : "MultiArray (Float16 1 × 1280 × 1 × 1500)",
"shortDescription" : "",
"shape" : "[1, 1280, 1, 1500]",
"name" : "encoder_output_embeds",
"type" : "MultiArray"
},
{
"hasShapeFlexibility" : "0",
"isOptional" : "0",
"dataType" : "Float16",
"formattedType" : "MultiArray (Float16 1 × 224)",
"shortDescription" : "",
"shape" : "[1, 224]",
"name" : "decoder_key_padding_mask",
"type" : "MultiArray"
}
],
"generatedClassName" : "TextDecoder",
"method" : "predict"
}
]