A high-performance inference engine for AI models
import Foundation
import Uzu
enum ReplyStatsFormat {
static let placeholder = "—"
static func energy(_ joules: Double) -> String {
measurement.string(from: Measurement(value: joules, unit: UnitEnergy.joules))
}
static func energyPerToken(_ stats: ChatReplyJoulesPerToken) -> String {
switch stats {
case let .total(total):
"total \(formattedNumber(total)) J/tok"
case let .components(cpu, gpu, _, dram):
"\(formattedNumber(cpu + gpu + dram)) J/tok"
}
}
static func memory(_ bytes: Int64?) -> String {
guard let bytes else { return placeholder }
return byteCount.string(fromByteCount: max(bytes, 0))
}
private static func formattedNumber(_ value: Double) -> String {
number.string(from: NSNumber(value: value)) ?? placeholder
}
private static let number: NumberFormatter = {
let formatter = NumberFormatter()
formatter.numberStyle = .decimal
formatter.minimumFractionDigits = 2
formatter.maximumFractionDigits = 2
return formatter
}()
private static let measurement: MeasurementFormatter = {
let formatter = MeasurementFormatter()
formatter.unitOptions = .providedUnit
formatter.numberFormatter = number
return formatter
}()
private static let byteCount: ByteCountFormatter = {
let formatter = ByteCountFormatter()
formatter.countStyle = .memory
formatter.allowedUnits = [.useMB, .useGB]
return formatter
}()
}