diff --git a/LLama.Examples/Assets/chat-with-kunkun-chinese.json b/LLama.Examples/Assets/chat-with-kunkun-chinese.json new file mode 100644 index 00000000..30112327 --- /dev/null +++ b/LLama.Examples/Assets/chat-with-kunkun-chinese.json @@ -0,0 +1,24 @@ +{ + "messages": [ + { + "author_role": "System", + "content": "������һ������û��ĶԻ��������������һ���ڸ����涼ӵ�зḻ�������������dz����ڻش��û�������Ͱ����û���?" + }, + { + "author_role": "User", + "content": "��ã�������?" + }, + { + "author_role": "Assistant", + "content": "��ã���ʲô���ܰ��������" + }, + { + "author_role": "User", + "content": "�й����׶����������У�" + }, + { + "author_role": "Assistant", + "content": "��������˭��" + } + ] +} diff --git a/LLama.Examples/Examples/ChatChineseGB2312.cs b/LLama.Examples/Examples/ChatChineseGB2312.cs index ff27b962..d250a454 100644 --- a/LLama.Examples/Examples/ChatChineseGB2312.cs +++ b/LLama.Examples/Examples/ChatChineseGB2312.cs @@ -1,69 +1,124 @@ -using System; -using System.Collections.Generic; -using System.Linq; -using System.Text; -using System.Threading.Tasks; +using System.Text; using LLama.Common; -namespace LLama.Examples.Examples +namespace LLama.Examples.Examples; + +public class ChatChineseGB2312 { - public class ChatChineseGB2312 + private static string ConvertEncoding(string input, Encoding original, Encoding target) + { + byte[] bytes = original.GetBytes(input); + var convertedBytes = Encoding.Convert(original, target, bytes); + return target.GetString(convertedBytes); + } + + public static async Task Run() { - private static string ConvertFromEncodingToAnother(string input, Encoding original, Encoding target) + // Register provider for GB2312 encoding + Encoding.RegisterProvider(CodePagesEncodingProvider.Instance); + + Console.ForegroundColor = ConsoleColor.Yellow; + Console.WriteLine("This example shows how to use Chinese with gb2312 encoding, which is common in windows. It's recommended" + + " to use https://huggingface.co/hfl/chinese-alpaca-2-7b-gguf/blob/main/ggml-model-q5_0.gguf, which has been verified by LLamaSharp developers."); + Console.ForegroundColor = ConsoleColor.White; + + Console.Write("Please input your model path: "); + var modelPath = Console.ReadLine(); + + var parameters = new ModelParams(modelPath) + { + ContextSize = 1024, + Seed = 1337, + GpuLayerCount = 5, + Encoding = Encoding.UTF8 + }; + using var model = LLamaWeights.LoadFromFile(parameters); + using var context = model.CreateContext(parameters); + var executor = new InteractiveExecutor(context); + + ChatSession session; + if (Directory.Exists("Assets/chat-with-kunkun-chinese")) + { + Console.ForegroundColor = ConsoleColor.Yellow; + Console.WriteLine("Loading session from disk."); + Console.ForegroundColor = ConsoleColor.White; + + session = new ChatSession(executor); + session.LoadSession("Assets/chat-with-kunkun-chinese"); + } + else { - byte[] bytes = original.GetBytes(input); - var convertedBytes = Encoding.Convert(original, target, bytes); - return target.GetString(convertedBytes); + var chatHistoryJson = File.ReadAllText("Assets/chat-with-kunkun-chinese.json"); + ChatHistory chatHistory = ChatHistory.FromJson(chatHistoryJson) ?? new ChatHistory(); + + session = new ChatSession(executor, chatHistory); } - public static async Task Run() + session + .WithHistoryTransform(new LLamaTransforms.DefaultHistoryTransform("用户")) + .WithOutputTransform(new LLamaTransforms.KeywordTextOutputStreamTransform( + // User and Assistant in Chinese (User is: 用户, Assistant is: 坤坤) + new string[] { "用户：", "坤坤：" }, + redundancyLength: 8)); + + InferenceParams inferenceParams = new InferenceParams() { - Encoding.RegisterProvider(CodePagesEncodingProvider.Instance); // Register gb2312 encoding - Console.Write("Please input your model path: "); - var modelPath = Console.ReadLine(); - var prompt = File.ReadAllText("Assets/chat-with-kunkun-chinese.txt", encoding: Encoding.GetEncoding("gb2312")).Trim(); - prompt = ConvertFromEncodingToAnother(prompt, Encoding.GetEncoding("gb2312"), Encoding.UTF8); + Temperature = 0.9f, + AntiPrompts = new List { "用户：" } + }; - var parameters = new ModelParams(modelPath) - { - ContextSize = 1024, - Seed = 1337, - GpuLayerCount = 20, - Encoding = Encoding.UTF8 - }; - using var model = LLamaWeights.LoadFromFile(parameters); - using var context = model.CreateContext(parameters); - var executor = new InteractiveExecutor(context); + Console.ForegroundColor = ConsoleColor.Yellow; + Console.WriteLine("The chat session has started."); - var session = new ChatSession(executor).WithHistoryTransform(new LLamaTransforms.DefaultHistoryTransform("用户")); + // show the prompt + Console.ForegroundColor = ConsoleColor.Green; + string userInput = Console.ReadLine() ?? ""; - Console.ForegroundColor = ConsoleColor.Yellow; - Console.WriteLine("This example shows how to use Chinese with gb2312 encoding, which is common in windows. It's recommended" + - " to use https://huggingface.co/hfl/chinese-alpaca-2-7b-gguf/blob/main/ggml-model-q5_0.gguf, which has been verified by LLamaSharp developers."); - Console.ForegroundColor = ConsoleColor.White; + while (userInput != "exit") + { + // Convert the encoding from gb2312 to utf8 for the language model + // and later saving to the history json file. + userInput = ConvertEncoding(userInput, Encoding.GetEncoding("gb2312"), Encoding.UTF8); - // show the prompt - Console.Write(prompt); - while (true) + if (userInput == "save") { - await foreach (var text in session.ChatAsync(prompt, new InferenceParams() + session.SaveSession("Assets/chat-with-kunkun-chinese"); + Console.ForegroundColor = ConsoleColor.Yellow; + Console.WriteLine("Session saved."); + } + else if (userInput == "regenerate") + { + Console.ForegroundColor = ConsoleColor.Yellow; + Console.WriteLine("Regenerating last response ..."); + + await foreach ( + var text + in session.RegenerateAssistantMessageAsync( + inferenceParams)) { - Temperature = 0.3f, - TopK = 5, - TopP = 0.85f, - AntiPrompts = new List { "用户：" }, - MaxTokens = 2048, - RepeatPenalty = 1.05f - })) + Console.ForegroundColor = ConsoleColor.White; + + // Convert the encoding from utf8 to gb2312 for the console output. + Console.Write(ConvertEncoding(text, Encoding.UTF8, Encoding.GetEncoding("gb2312"))); + } + } + else + { + await foreach ( + var text + in session.ChatAsync( + new ChatHistory.Message(AuthorRole.User, userInput), + inferenceParams)) { - //Console.Write(text); - Console.Write(ConvertFromEncodingToAnother(text, Encoding.UTF8, Encoding.GetEncoding("gb2312"))); + Console.ForegroundColor = ConsoleColor.White; + Console.Write(text); } - - Console.ForegroundColor = ConsoleColor.Green; - prompt = Console.ReadLine(); - Console.ForegroundColor = ConsoleColor.White; } + + Console.ForegroundColor = ConsoleColor.Green; + userInput = Console.ReadLine() ?? ""; + + Console.ForegroundColor = ConsoleColor.White; } } } diff --git a/LLama.Examples/Examples/Runner.cs b/LLama.Examples/Examples/Runner.cs index 0a37dcba..3d9858e1 100644 --- a/LLama.Examples/Examples/Runner.cs +++ b/LLama.Examples/Examples/Runner.cs @@ -9,6 +9,7 @@ public class Runner { "Run a chat session with history.", ChatSessionWithHistory.Run }, { "Run a chat session without stripping the role names.", ChatSessionWithRoleName.Run }, { "Run a chat session with the role names stripped.", ChatSessionStripRoleName.Run }, + { "Run a chat session in Chinese GB2312 encoding", ChatChineseGB2312.Run }, { "Interactive mode chat by using executor.", InteractiveModeExecute.Run }, { "Instruct mode chat by using executor.", InstructModeExecute.Run }, { "Stateless mode chat by using executor.", StatelessModeExecute.Run }, @@ -24,7 +25,6 @@ public class Runner { "Coding Assistant.", CodingAssistant.Run }, { "Batch Decoding.", BatchedDecoding.Run }, { "SK Kernel Memory.", KernelMemory.Run }, - { "Chinese gb2312 chat", ChatChineseGB2312.Run }, { "Exit", async () => Environment.Exit(0) } };