llama.swiftui : UX improvements

examples/llama.swiftui/llama.cpp.swift/LibLlama.swift

@@ -30,7 +30,7 @@ actor LlamaContext {
     /// This variable is used to store temporarily invalid cchars
     private var temporary_invalid_cchars: [CChar]
 
-    var n_len: Int32 = 512
+    var n_len: Int32 = 64
     var n_cur: Int32 = 0
 
     var n_decode: Int32 = 0
@@ -241,6 +241,8 @@ actor LlamaContext {
 
         let t_tg_end = ggml_time_us()
 
+        llama_kv_cache_clear(context)
+
         let t_pp = Double(t_pp_end - t_pp_start) / 1000000.0
         let t_tg = Double(t_tg_end - t_tg_start) / 1000000.0
 
@@ -254,11 +256,12 @@ actor LlamaContext {
     func clear() {
         tokens_list.removeAll()
         temporary_invalid_cchars.removeAll()
+        llama_kv_cache_clear(context)
     }
 
     private func tokenize(text: String, add_bos: Bool) -> [llama_token] {
         let utf8Count = text.utf8.count
-        let n_tokens = utf8Count + (add_bos ? 1 : 0)
+        let n_tokens = utf8Count + (add_bos ? 1 : 0) + 1
         let tokens = UnsafeMutablePointer<llama_token>.allocate(capacity: n_tokens)
         let tokenCount = llama_tokenize(model, text, Int32(utf8Count), tokens, Int32(n_tokens), add_bos, false)
 

examples/llama.swiftui/llama.swiftui/Models/LlamaState.swift

@@ -34,7 +34,6 @@ class LlamaState: ObservableObject {
             return
         }
 
-        messageLog += "Attempting to complete text...\n"
         await llamaContext.completion_init(text: text)
         messageLog += "\(text)"
 
@@ -58,4 +57,13 @@ class LlamaState: ObservableObject {
         let result = await llamaContext.bench()
         messageLog += "\(result)"
     }
+
+    func clear() async {
+        guard let llamaContext else {
+            return
+        }
+
+        await llamaContext.clear()
+        messageLog = ""
+    }
 }

examples/llama.swiftui/llama.swiftui/UI/ContentView.swift

@@ -23,22 +23,26 @@ struct ContentView: View {
 
     var body: some View {
         VStack {
-            // automatically scroll to bottom of text view
             ScrollView(.vertical, showsIndicators: true) {
                 Text(llamaState.messageLog)
+                .font(.system(size: 12))
+                .frame(maxWidth: .infinity, alignment: .leading)
+                .padding()
+                .onTapGesture {
+                    UIApplication.shared.sendAction(#selector(UIResponder.resignFirstResponder), to: nil, from: nil, for: nil)
+                }
             }
 
             TextEditor(text: $multiLineText)
-                .frame(height: 200)
+                .frame(height: 80)
                 .padding()
                 .border(Color.gray, width: 0.5)
 
-            // add two buttons "Send" and "Bench" next to each other
             HStack {
                 Button("Send") {
                     sendText()
                 }
-                .padding()
+                .padding(8)
                 .background(Color.blue)
                 .foregroundColor(.white)
                 .cornerRadius(8)
@@ -46,7 +50,15 @@ struct ContentView: View {
                 Button("Bench") {
                     bench()
                 }
-                .padding()
+                .padding(8)
+                .background(Color.blue)
+                .foregroundColor(.white)
+                .cornerRadius(8)
+
+                Button("Clear") {
+                    clear()
+                }
+                .padding(8)
                 .background(Color.blue)
                 .foregroundColor(.white)
                 .cornerRadius(8)
@@ -55,20 +67,27 @@ struct ContentView: View {
             VStack {
                 DownloadButton(
                     llamaState: llamaState,
-                    modelName: "TheBloke / TinyLlama-1.1B-1T-OpenOrca-GGUF (Q4_0)",
+                    modelName: "TinyLlama-1.1B (Q4_0)",
                     modelUrl: "https://huggingface.co/TheBloke/TinyLlama-1.1B-1T-OpenOrca-GGUF/resolve/main/tinyllama-1.1b-1t-openorca.Q4_0.gguf?download=true",
                     filename: "tinyllama-1.1b-1t-openorca.Q4_0.gguf"
                 )
+                .font(.system(size: 12))
+                .padding(.top, 4)
+
                 DownloadButton(
                     llamaState: llamaState,
-                    modelName: "TheBloke / TinyLlama-1.1B-1T-OpenOrca-GGUF (Q8_0)",
+                    modelName: "TinyLlama-1.1B (Q8_0)",
                     modelUrl: "https://huggingface.co/TheBloke/TinyLlama-1.1B-1T-OpenOrca-GGUF/resolve/main/tinyllama-1.1b-1t-openorca.Q8_0.gguf?download=true",
                     filename: "tinyllama-1.1b-1t-openorca.Q8_0.gguf"
                 )
+                .font(.system(size: 12))
+
                 Button("Clear downloaded models") {
                     ContentView.cleanupModelCaches()
                     llamaState.cacheCleared = true
                 }
+                .padding(8)
+                .font(.system(size: 12))
             }
         }
         .padding()
@@ -86,6 +105,12 @@ struct ContentView: View {
             await llamaState.bench()
         }
     }
+
+    func clear() {
+        Task {
+            await llamaState.clear()
+        }
+    }
 }
 
 //#Preview {

examples/llama.swiftui/llama.swiftui/UI/DownloadButton.swift

@@ -31,6 +31,7 @@ struct DownloadButton: View {
 
     private func download() {
         status = "downloading"
+        print("Downloading model \(modelName) from \(modelUrl)")
         downloadTask = URLSession.shared.dataTask(with: URL(string: modelUrl)!) { data, response, error in
             if let error = error {
                 print("Error: \(error.localizedDescription)")
@@ -44,12 +45,12 @@ struct DownloadButton: View {
 
             if let data = data {
                 do {
+                    print("Writing to \(filename)")
                     let fileURL = DownloadButton.getFileURL(filename: filename)
                     try data.write(to: fileURL)
 
                     llamaState.cacheCleared = false
 
-                    print("Downloaded model \(modelName) to \(fileURL)")
                     status = "downloaded"
                     try llamaState.loadModel(modelUrl: fileURL)
                 } catch let err {