llm improvements

- document the tokenizer used (https://github.com/huggingface/swift-transformers)
- provide a hook for tokenizer configuration, prompt augmentation
	- this isn't as rich as the python equivalents but it helps a little

Libraries/LLM/Load.swift

@@ -14,13 +14,13 @@ struct LLMError: Error {
 
 /// Load and return the model and tokenizer
 public func load(
-    hub: HubApi = HubApi(), name: String, progressHandler: @escaping (Progress) -> Void = { _ in }
+    hub: HubApi = HubApi(), configuration: ModelConfiguration, progressHandler: @escaping (Progress) -> Void = { _ in }
 ) async throws -> (LLMModel, Tokenizer) {
     // note: this doesn't have a way to pass the HubApi
-    let tokenizer = try await loadTokenizer(name: name)
+    let tokenizer = try await loadTokenizer(configuration: configuration)
 
     // download the model weights and config
-    let repo = Hub.Repo(id: name)
+    let repo = Hub.Repo(id: configuration.id)
     let modelFiles = ["config.json", "*.safetensors"]
     let modelDirectory = try await hub.snapshot(
         from: repo, matching: modelFiles, progressHandler: progressHandler)

Libraries/LLM/Models.swift

@@ -0,0 +1,107 @@
+// Copyright © 2024 Apple Inc.
+
+import Foundation
+
+/// Registry of models and and any overrides that go with them, e.g. prompt augmentation.
+/// If asked for an unknown configuration this will use the model/tokenizer as-is.
+///
+/// The python tokenizers have a very rich set of implementations and configuration.  The
+/// swift-tokenizers code handles a good chunk of that and this is a place to augment that
+/// implementation, if needed.
+public struct ModelConfiguration {
+    public let id: String
+    
+    /// overrides for TokenizerModel/knownTokenizers -- useful before swift-transformers is updated
+    public let overrideTokenizer: String?
+    
+    /// custom preparation logic for the prompt.  custom tokenizers provide more capability, but this
+    /// allows some minor formtting changes, e.g. wrapping the user input in the expected prompt
+    /// format
+    private let preparePrompt: ((String) -> String)?
+    
+    public init(id: String, overrideTokenizer: String? = nil, preparePrompt: ((String) -> String)? = nil) {
+        self.id = id
+        self.overrideTokenizer = overrideTokenizer
+        self.preparePrompt = preparePrompt
+    }
+    
+    public func prepare(prompt: String) -> String {
+        preparePrompt?(prompt) ?? prompt
+    }
+    
+    public static var registry = [String:ModelConfiguration]()
+    
+    public static func register(configurations: [ModelConfiguration]) {
+        bootstrap()
+        
+        for c in configurations {
+            registry[c.id] = c
+        }
+    }
+    
+    public static func configuration(id: String) -> ModelConfiguration {
+        bootstrap()
+
+        if let c = registry[id] {
+            return c
+        } else {
+            return ModelConfiguration(id: id)
+        }
+    }
+}
+
+extension ModelConfiguration {
+    
+    static let mistral7B4bit = ModelConfiguration(id: "mlx-community/Mistral-7B-v0.1-hf-4bit-mlx")
+
+    static let codeLlama13b4bit = ModelConfiguration(
+        id: "mlx-community/CodeLlama-13b-Instruct-hf-4bit-MLX",
+        overrideTokenizer: "PreTrainedTokenizer")
+    { prompt in
+        // given the prompt: func sortArray(_ array: [Int]) -> String { <FILL_ME> }
+        // the python code produces this (via its custom tokenizer):
+        // <PRE> func sortArray(_ array: [Int]) -> String {  <SUF> } <MID>
+        
+        "<PRE> " +
+        prompt.replacingOccurrences(of: "<FILL_ME>", with: "<SUF>") +
+        " <MID>"
+    }
+    
+    static let phi4bit = ModelConfiguration(id: "mlx-community/phi-2-hf-4bit-mlx") { prompt in
+        "Instruct: \(prompt). Output: "
+    }
+    
+    static let gemma2bQuantized = ModelConfiguration(
+        id: "mlx-community/quantized-gemma-2b-it",
+        overrideTokenizer: "PreTrainedTokenizer") { prompt in
+            "<start_of_turn>user \(prompt)<end_of_turn><start_of_turn>model"
+        }
+
+    private enum BootstrapState {
+        case idle
+        case bootstrapping
+        case bootstrapped
+    }
+    
+    static private var bootstrapState = BootstrapState.idle
+    
+    static func bootstrap() {
+        switch bootstrapState {
+        case .idle:
+            bootstrapState = .bootstrapping
+            register(configurations: [
+                mistral7B4bit,
+                codeLlama13b4bit,
+                phi4bit,
+                gemma2bQuantized,
+            ])
+            bootstrapState = .bootstrapped
+
+        case .bootstrapping:
+            break
+            
+        case .bootstrapped:
+            break
+        }
+    }
+}

Libraries/LLM/README.md

@@ -4,9 +4,22 @@ This is a port of several models from:
 
 - https://github.com/ml-explore/mlx-examples/blob/main/llms/mlx_lm/models/
 
-You can use this to load models from huggingface, e.g.:
+using the Hugging Face swift transformers package to provide tokenization:
 
-- https://huggingface.co/mlx-community/Mistral-7B-v0.1-hf-4bit-mlx
+https://github.com/huggingface/swift-transformers
+
+The [Models.swift](Models.swift) provides minor overrides and customization --
+if you require overrides for the tokenizer or prompt customizations they can be
+added there.
+
+This is set up to load models from Hugging Face, e.g. https://huggingface.co/mlx-community
+
+The following models have been tried:
+
+- mlx-community/Mistral-7B-v0.1-hf-4bit-mlx
+- mlx-community/CodeLlama-13b-Instruct-hf-4bit-MLX
+- mlx-community/phi-2-hf-4bit-mlx
+- mlx-community/quantized-gemma-2b-it
 
 Currently supported model types are:
 

Libraries/LLM/Tokenizer.swift

@@ -49,9 +49,9 @@ public struct Tokenizer: Tokenizers.Tokenizer {
 
 }
 
-public func loadTokenizer(name: String) async throws -> Tokenizer {
+public func loadTokenizer(configuration: ModelConfiguration) async throws -> Tokenizer {
     // from AutoTokenizer.from() -- this lets us override parts of the configuration
-    let config = LanguageModelConfigurationFromHub(modelName: name)
+    let config = LanguageModelConfigurationFromHub(modelName: configuration.id)
     guard var tokenizerConfig = try await config.tokenizerConfig else {
         throw LLMError(message: "missing config")
     }

Tools/llm-tool/LLMTool.swift

@@ -25,7 +25,7 @@ struct SyncGenerator: AsyncParsableCommand {
     var model: String = "mlx-community/Mistral-7B-v0.1-hf-4bit-mlx"
 
     @Option(name: .shortAndLong, help: "The message to be processed by the model")
-    var prompt = "compare swift and python"
+    var prompt = "compare python and swift"
 
     @Option(name: .shortAndLong, help: "Maximum number of tokens to generate")
     var maxTokens = 100
@@ -40,22 +40,24 @@ struct SyncGenerator: AsyncParsableCommand {
     func run() async throws {
         MLXRandom.seed(seed)
 
-        let (model, tokenizer) = try await load(name: model)
+        let modelConfiguration = ModelConfiguration.configuration(id: model)
+        let (model, tokenizer) = try await load(configuration: modelConfiguration)
+        
+        let prompt = modelConfiguration.prepare(prompt: self.prompt)
+        let promptTokens = tokenizer.encode(text: prompt)
 
         print("Starting generation ...")
-        print(prompt, terminator: "")
+        print(self.prompt, terminator: "")
 
         var start = Date.timeIntervalSinceReferenceDate
         var promptTime: TimeInterval = 0
 
-        let prompt = MLXArray(tokenizer.encode(text: prompt))
-
         // collect the tokens and keep track of how much of the string
         // we have printed already
         var tokens = [Int]()
         var printed = 0
 
-        for token in TokenIterator(prompt: prompt, model: model, temp: temperature) {
+        for token in TokenIterator(prompt: MLXArray(promptTokens), model: model, temp: temperature) {
             if tokens.isEmpty {
                 eval(token)
                 let now = Date.timeIntervalSinceReferenceDate
@@ -90,7 +92,7 @@ struct SyncGenerator: AsyncParsableCommand {
 
         print(
             """
-            Prompt Tokens per second:     \((Double(prompt.size) / promptTime).formatted())
+            Prompt Tokens per second:     \((Double(promptTokens.count) / promptTime).formatted())
             Generation tokens per second: \((Double(tokens.count - 1) / generateTime).formatted())
             """)
     }
@@ -111,7 +113,7 @@ struct AsyncGenerator: AsyncParsableCommand {
     var model: String = "mlx-community/Mistral-7B-v0.1-hf-4bit-mlx"
 
     @Option(name: .shortAndLong, help: "The message to be processed by the model")
-    var prompt = "compare swift and python"
+    var prompt = "compare python and swift"
 
     @Option(name: .shortAndLong, help: "Maximum number of tokens to generate")
     var maxTokens = 100
@@ -126,22 +128,24 @@ struct AsyncGenerator: AsyncParsableCommand {
     func run() async throws {
         MLXRandom.seed(seed)
 
-        let (model, tokenizer) = try await load(name: model)
+        let modelConfiguration = ModelConfiguration.configuration(id: model)
+        let (model, tokenizer) = try await load(configuration: modelConfiguration)
+        
+        let prompt = modelConfiguration.prepare(prompt: self.prompt)
+        let promptTokens = tokenizer.encode(text: prompt)
 
         print("Starting generation ...")
-        print(prompt, terminator: "")
+        print(self.prompt, terminator: "")
 
         var start = Date.timeIntervalSinceReferenceDate
         var promptTime: TimeInterval = 0
 
-        let prompt = MLXArray(tokenizer.encode(text: prompt))
-
         // collect the tokens and keep track of how much of the string
         // we have printed already
         var tokens = [Int]()
         var printed = 0
 
-        let (task, channel) = generate(prompt: prompt, model: model, temp: temperature)
+        let (task, channel) = generate(prompt: MLXArray(promptTokens), model: model, temp: temperature)
 
         for await token in channel {
             if tokens.isEmpty {
@@ -179,7 +183,7 @@ struct AsyncGenerator: AsyncParsableCommand {
 
         print(
             """
-            Prompt Tokens per second:     \((Double(prompt.size) / promptTime).formatted())
+            Prompt Tokens per second:     \((Double(promptTokens.count) / promptTime).formatted())
             Generation tokens per second: \((Double(tokens.count - 1) / generateTime).formatted())
             """)
 

mlx-swift-examples.xcodeproj/project.pbxproj

@@ -36,6 +36,7 @@
 		C3932D572B6A060B00A81055 /* MNIST.swift in Sources */ = {isa = PBXBuildFile; fileRef = C3932D562B6A060B00A81055 /* MNIST.swift */; };
 		C3932D592B6A0BE400A81055 /* Random.swift in Sources */ = {isa = PBXBuildFile; fileRef = C3932D582B6A0BE400A81055 /* Random.swift */; };
 		C397C59C2B62C6D0004B084D /* ArgumentParser in Frameworks */ = {isa = PBXBuildFile; productRef = C397C59B2B62C6D0004B084D /* ArgumentParser */; };
+		C3A8B3AC2B9283150002EFB8 /* Models.swift in Sources */ = {isa = PBXBuildFile; fileRef = C3A8B3AB2B9283150002EFB8 /* Models.swift */; };
 		C3E786AB2B8D1AEC0004D037 /* Evaluate.swift in Sources */ = {isa = PBXBuildFile; fileRef = C3E786AA2B8D1AEC0004D037 /* Evaluate.swift */; };
 		C3E786AD2B8D4AF50004D037 /* Tokenizer.swift in Sources */ = {isa = PBXBuildFile; fileRef = C3E786AC2B8D4AF50004D037 /* Tokenizer.swift */; };
 		C3FBCB212B8520B80007E490 /* MLX in Frameworks */ = {isa = PBXBuildFile; productRef = C3FBCB202B8520B80007E490 /* MLX */; };
@@ -152,6 +153,7 @@
 		C3932D562B6A060B00A81055 /* MNIST.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = MNIST.swift; sourceTree = "<group>"; };
 		C3932D582B6A0BE400A81055 /* Random.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = Random.swift; sourceTree = "<group>"; };
 		C397C58B2B62C6A9004B084D /* llm-tool */ = {isa = PBXFileReference; explicitFileType = "compiled.mach-o.executable"; includeInIndex = 0; path = "llm-tool"; sourceTree = BUILT_PRODUCTS_DIR; };
+		C3A8B3AB2B9283150002EFB8 /* Models.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = Models.swift; sourceTree = "<group>"; };
 		C3C3240B2B6CA689007D2D9A /* README.md */ = {isa = PBXFileReference; lastKnownFileType = net.daringfireball.markdown; path = README.md; sourceTree = "<group>"; };
 		C3C3240C2B6CA792007D2D9A /* README.md */ = {isa = PBXFileReference; lastKnownFileType = net.daringfireball.markdown; path = README.md; sourceTree = "<group>"; };
 		C3E786AA2B8D1AEC0004D037 /* Evaluate.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = Evaluate.swift; sourceTree = "<group>"; };
@@ -267,6 +269,7 @@
 			isa = PBXGroup;
 			children = (
 				C34E48EF2B696E6500FCB841 /* Configuration.swift */,
+				C3A8B3AB2B9283150002EFB8 /* Models.swift */,
 				C34E48EE2B696E6500FCB841 /* Llama.swift */,
 				C38935E22B86C0FE0037B833 /* Gemma.swift */,
 				C38935C72B869C7A0037B833 /* LLM.h */,
@@ -614,6 +617,7 @@
 				C38935DF2B869DD00037B833 /* Phi.swift in Sources */,
 				C38935CE2B869C870037B833 /* Load.swift in Sources */,
 				C3E786AD2B8D4AF50004D037 /* Tokenizer.swift in Sources */,
+				C3A8B3AC2B9283150002EFB8 /* Models.swift in Sources */,
 				C3E786AB2B8D1AEC0004D037 /* Evaluate.swift in Sources */,
 				C38935CC2B869C870037B833 /* Llama.swift in Sources */,
 			);

mlx-swift-examples.xcodeproj/project.xcworkspace/xcshareddata/swiftpm/Package.resolved

@@ -15,7 +15,7 @@
       "location" : "https://github.com/ml-explore/mlx-swift",
       "state" : {
         "branch" : "main",
-        "revision" : "fbe215ae29ec286bdc66f4b3423e3eea1b3ef2fc"
+        "revision" : "83bd19f9fe93d77d9f89981eeead2d6d190afdba"
       }
     },
     {

mlx-swift-examples.xcodeproj/xcshareddata/xcschemes/llm-tool.xcscheme

@@ -0,0 +1,100 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<Scheme
+   LastUpgradeVersion = "1520"
+   version = "1.7">
+   <BuildAction
+      parallelizeBuildables = "YES"
+      buildImplicitDependencies = "YES">
+      <BuildActionEntries>
+         <BuildActionEntry
+            buildForTesting = "YES"
+            buildForRunning = "YES"
+            buildForProfiling = "YES"
+            buildForArchiving = "YES"
+            buildForAnalyzing = "YES">
+            <BuildableReference
+               BuildableIdentifier = "primary"
+               BlueprintIdentifier = "C397C58A2B62C6A9004B084D"
+               BuildableName = "llm-tool"
+               BlueprintName = "llm-tool"
+               ReferencedContainer = "container:mlx-swift-examples.xcodeproj">
+            </BuildableReference>
+         </BuildActionEntry>
+      </BuildActionEntries>
+   </BuildAction>
+   <TestAction
+      buildConfiguration = "Debug"
+      selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
+      selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
+      shouldUseLaunchSchemeArgsEnv = "YES"
+      shouldAutocreateTestPlan = "YES">
+   </TestAction>
+   <LaunchAction
+      buildConfiguration = "Release"
+      selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
+      selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
+      launchStyle = "0"
+      useCustomWorkingDirectory = "NO"
+      ignoresPersistentStateOnLaunch = "NO"
+      debugDocumentVersioning = "YES"
+      debugServiceExtension = "internal"
+      allowLocationSimulation = "YES"
+      viewDebuggingEnabled = "No">
+      <BuildableProductRunnable
+         runnableDebuggingMode = "0">
+         <BuildableReference
+            BuildableIdentifier = "primary"
+            BlueprintIdentifier = "C397C58A2B62C6A9004B084D"
+            BuildableName = "llm-tool"
+            BlueprintName = "llm-tool"
+            ReferencedContainer = "container:mlx-swift-examples.xcodeproj">
+         </BuildableReference>
+      </BuildableProductRunnable>
+      <CommandLineArguments>
+         <CommandLineArgument
+            argument = "--model mlx-community/CodeLlama-13b-Instruct-hf-4bit-MLX"
+            isEnabled = "NO">
+         </CommandLineArgument>
+         <CommandLineArgument
+            argument = "--prompt &apos;func sortArray(_ array: [Int]) -&gt; String { &lt;FILL_ME&gt; }&apos;"
+            isEnabled = "NO">
+         </CommandLineArgument>
+         <CommandLineArgument
+            argument = "--model mlx-community/Mistral-7B-v0.1-hf-4bit-mlx"
+            isEnabled = "NO">
+         </CommandLineArgument>
+         <CommandLineArgument
+            argument = "--model mlx-community/quantized-gemma-2b-it"
+            isEnabled = "NO">
+         </CommandLineArgument>
+         <CommandLineArgument
+            argument = "--model mlx-community/phi-2-hf-4bit-mlx"
+            isEnabled = "YES">
+         </CommandLineArgument>
+      </CommandLineArguments>
+   </LaunchAction>
+   <ProfileAction
+      buildConfiguration = "Release"
+      shouldUseLaunchSchemeArgsEnv = "YES"
+      savedToolIdentifier = ""
+      useCustomWorkingDirectory = "NO"
+      debugDocumentVersioning = "YES">
+      <BuildableProductRunnable
+         runnableDebuggingMode = "0">
+         <BuildableReference
+            BuildableIdentifier = "primary"
+            BlueprintIdentifier = "C397C58A2B62C6A9004B084D"
+            BuildableName = "llm-tool"
+            BlueprintName = "llm-tool"
+            ReferencedContainer = "container:mlx-swift-examples.xcodeproj">
+         </BuildableReference>
+      </BuildableProductRunnable>
+   </ProfileAction>
+   <AnalyzeAction
+      buildConfiguration = "Debug">
+   </AnalyzeAction>
+   <ArchiveAction
+      buildConfiguration = "Release"
+      revealArchiveInOrganizer = "YES">
+   </ArchiveAction>
+</Scheme>