Merge pull request #22 from Tabing010102/master

更新GalTransl prompt、修复Sakura 0.8/0.9
fkiliver · Oct 26, 2024 · 1fe2c77 · 1fe2c77
2 parents e458bb8 + 857ebad
commit 1fe2c77
Show file tree

Hide file tree

Showing 5 changed files with 87 additions and 98 deletions.
diff --git a/README.md b/README.md
@@ -6,7 +6,7 @@
 
 # 介绍
 这是一个基于XUnity.AutoTranslator和Sakura模型的Unity游戏本地翻译器,能够提供高质量离线日文翻译  
-建议使用[Galtransl-v2.6翻译模型](https://github.com/SakuraLLM/Sakura-13B-Galgame)，当前支持版本为Sakura v0.8/v0.9/v0.10/v1.0,GalTrans2.6
+建议使用[Galtransl-v2.6翻译模型](https://huggingface.co/SakuraLLM/GalTransl-7B-v2.6)，当前支持版本为Sakura v0.8/v0.9/v0.10/v1.0,GalTrans2.6
 
 ## TODO
 - [ ] 添加退化检测（搁置，较新的模型基本不需要）
@@ -60,7 +60,7 @@ ModelName=Sakura
 ModelVersion=1.0
 MaxConcurrency=2
 UseDict=True
-DictMode=Full
+DictMode=Partial
 Dict={"アイリス":["艾莉斯","女"]}
 ```
 
@@ -69,8 +69,8 @@ Dict={"アイリス":["艾莉斯","女"]}
 | ModelName  | ModelVersion | TranslationModel       |
 |------------|--------------|------------------------|
 | Sakura     | 0.8          | Sakura 0.8             |
-| Sakura     | 0.9          | Sakura/Sakura32B 0.9*             |
-| Sakura     | 0.10         | Sakura 0.10pre1            |
+| Sakura     | 0.9          | Sakura 7B/14B/32B 0.9  |
+| Sakura     | 0.10         | Sakura 0.10            |
 | Sakura     | 1.0          | Sakura 1.0             |
 | Sakura     | *            | Sakura 1.0 (默认)      |
 | Sakura32B  | 0.10         | Sakura32B 0.10         |
@@ -79,7 +79,9 @@ Dict={"アイリス":["艾莉斯","女"]}
 | GalTransl  | *            | GalTransl 2.6 (默认)   |
 | *          | *            | Sakura 1.0 (默认)      |
 
-模型相关默认值设置为Sakura 1.0，注意Sakura32B 0.9*需将`ModelName`设置为`Sakura`
+模型相关默认值设置为Sakura 1.0，注意Sakura 7B/14B/32B 0.9系列模型需将`ModelName`设置为`Sakura`  
+其中Sakura 0.8/0.9模型需要将`Endpoint`设置为completion api（例：`http://127.0.0.1:8080/completion`）  
+Sakura 0.10/1.0和GalTransl模型需要将`Endpoint`设置为chat completions api（例：`http://127.0.0.1:8080/v1/chat/completions`）  
 
 ### 字典
 #### 字典配置项

diff --git a/SakuraTranslator/SakuraTranslate.csproj b/SakuraTranslator/SakuraTranslate.csproj
@@ -55,6 +55,7 @@
     </Reference>
   </ItemGroup>
   <ItemGroup>
+    <Compile Include="SakuraUtil.cs" />
     <Compile Include="TranslationModel.cs" />
     <Compile Include="SakuraTranslateEndpoint.cs" />
   </ItemGroup>

diff --git a/SakuraTranslator/SakuraTranslateEndpoint.cs b/SakuraTranslator/SakuraTranslateEndpoint.cs
@@ -12,8 +12,8 @@
 using XUnity.AutoTranslator.Plugin.Core.Endpoints;
 using XUnity.AutoTranslator.Plugin.Core.Utilities;
 
-[assembly: AssemblyVersion("0.3.4")]
-[assembly: AssemblyFileVersion("0.3.4")]
+[assembly: AssemblyVersion("0.3.5")]
+[assembly: AssemblyFileVersion("0.3.5")]
 
 namespace SakuraTranslate
 {
@@ -42,12 +42,12 @@ public partial class SakuraTranslateEndpoint : ITranslateEndpoint
 
         public void Initialize(IInitializationContext context)
         {
-            _endpoint = context.GetOrCreateSetting<string>("Sakura", "Endpoint", "http://127.0.0.1:5000/v1/chat/completions");
+            _endpoint = context.GetOrCreateSetting<string>("Sakura", "Endpoint", "http://127.0.0.1:8080/v1/chat/completions");
             _modelName = context.GetOrCreateSetting<string>("Sakura", "ModelName", "Sakura");
             _modelVersion = context.GetOrCreateSetting<string>("Sakura", "ModelVersion", "1.0");
             _modelType = GetTranslationModel(_modelName, _modelVersion);
             if (!int.TryParse(context.GetOrCreateSetting<string>("Sakura", "MaxConcurrency", "1"), out _maxConcurrency))
-            { 
+            {
                 _maxConcurrency = 1;
             }
             if (_maxConcurrency > ServicePointManager.DefaultConnectionLimit)
@@ -178,22 +178,17 @@ public IEnumerator Translate(ITranslationContext context)
             //var translatedText = responseText.Substring(startIndex, endIndex - startIndex);
 
             JObject jsonResponse = JObject.Parse(responseText);
-            string translatedText = jsonResponse["choices"]?[0]?["message"]?["content"]?.ToString();
-
-
-            //历史遗留
-            //if (translatedLine.EndsWith("<|im_end|>"))
-            //{
-            //    translatedLine = translatedLine.Substring(0, translatedLine.Length - "<|im_end|>".Length);
-            //}
-            //if (translatedLine.EndsWith("。") && !line.Trim().EndsWith("。"))
-            //{
-            //    translatedLine = translatedLine.Substring(0, translatedLine.Length - "。".Length);
-            //}
-            //if (translatedLine.EndsWith("。」") && !line.Trim().EndsWith("。」"))
-            //{
-            //    translatedLine = translatedLine.Substring(0, translatedLine.Length - "。」".Length) + "」";
-            //}
+            string translatedText;
+            if (IsOpenAIEndpoint(_modelType))
+            {
+                translatedText = jsonResponse["choices"]?[0]?["message"]?["content"]?.ToString();
+            }
+            else
+            {
+                translatedText = jsonResponse["content"]?.ToString();
+            }
+
+            translatedText = SakuraUtil.FixTranslationEnd(untranslatedText, translatedText);
 
             // 提交翻译文本
             context.Complete(translatedText);
@@ -372,78 +367,58 @@ private string MakeSakura32bPromptV0_10(string line)
 
         private string MakeSakuraPromptV1_0(string line)
         {
-            string messagesStr = string.Empty;
-            if (_useDict == false)
-            {
-                messagesStr = SerializePromptMessages(new List<PromptMessage>
+            var messages = new List<PromptMessage>
                 {
                     new PromptMessage
                     {
                         Role = "system",
                         Content = "你是一个轻小说翻译模型，可以流畅通顺地以日本轻小说的风格将日文翻译成简体中文，并联系上下文正确使用人称代词，不擅自添加原文中没有的代词。"
-                    },
-                    new PromptMessage
-                    {
-                        Role = "user",
-                        Content = $"将下面的日文文本翻译成中文：{line}"
                     }
-                });
+                };
+            string dictStr;
+            if (_useDict == false)
+            {
+                dictStr = string.Empty;
+            }
+            if (_dictMode == "Full")
+            {
+                dictStr = _fullDictStr;
             }
             else
             {
-                var messages = new List<PromptMessage>
-                {
-                    new PromptMessage
-                    {
-                        Role = "system",
-                        Content = "你是一个轻小说翻译模型，可以流畅通顺地以日本轻小说的风格将日文翻译成简体中文，并联系上下文正确使用人称代词，不擅自添加原文中没有的代词。"
-                    }
-                };
-                string dictStr;
-                if (_useDict == false)
-                {
-                    dictStr = string.Empty;
-                }
-                if (_dictMode == "Full")
+                var usedDict = _dict.Where(x => line.Contains(x.Key));
+                if (usedDict.Count() > 0)
                 {
-                    dictStr = _fullDictStr;
+                    var dictStrings = GetDictStringList(usedDict);
+                    dictStr = string.Join("\n", dictStrings.ToArray());
                 }
                 else
                 {
-                    var usedDict = _dict.Where(x => line.Contains(x.Key));
-                    if (usedDict.Count() > 0)
-                    {
-                        var dictStrings = GetDictStringList(usedDict);
-                        dictStr = string.Join("\n", dictStrings.ToArray());
-                    }
-                    else
-                    {
-                        dictStr = string.Empty;
-                    }
+                    dictStr = string.Empty;
                 }
+            }
 
-                if (_useDict == false)
+            if (_useDict == false)
+            {
+                // 如果术语表为空，直接构建翻译指令
+                messages.Add(new PromptMessage
                 {
-                    // 如果术语表为空，直接构建翻译指令
-                    messages.Add(new PromptMessage
-                    {
-                        Role = "user",
-                        Content = $"将下面的日文文本翻译成中文：{line}"
-                    });
-                }
-                else
-
+                    Role = "user",
+                    Content = $"将下面的日文文本翻译成中文：{line}"
+                });
+            }
+            else
+            {
+                messages.Add(new PromptMessage
                 {
-                    messages.Add(new PromptMessage
-                    {
-                        Role = "user",
-                        Content = $"根据以下术语表（可以为空）：\n{dictStr}\n" +
-                              $"将下面的日文文本根据对应关系和备注翻译成中文：{line}"
-                    });
-                }
-
-                messagesStr = SerializePromptMessages(messages);
+                    Role = "user",
+                    Content = $"根据以下术语表（可以为空）：\n{dictStr}\n" +
+                          $"将下面的日文文本根据对应关系和备注翻译成中文：{line}"
+                });
             }
+
+            var messagesStr = SerializePromptMessages(messages);
+
             return $"{{" +
                    $"\"model\": \"sukinishiro\"," +
                    $"\"messages\": " +
@@ -461,7 +436,6 @@ private string MakeSakuraPromptV1_0(string line)
 
         private string MakeGalTranslPromptV2_6(string line)
         {
-            string messagesStr = string.Empty;
             var messages = new List<PromptMessage>
                 {
                     new PromptMessage
@@ -493,26 +467,14 @@ private string MakeGalTranslPromptV2_6(string line)
                 }
             }
 
-            if (_useDict == false)
-            {
-                // 如果术语表为空，直接构建翻译指令
-                messages.Add(new PromptMessage
-                {
-                    Role = "user",
-                    Content = $"将下面的日文文本翻译成中文：{line}"
-                });
-            }
-            else
+            messages.Add(new PromptMessage
             {
-                messages.Add(new PromptMessage
-                {
-                    Role = "user",
-                    Content = $"参考以下术语表（可为空，格式为src->dst #备注）：\n{dictStr}\n" +
-                                $"根据上述术语表的对应关系和备注，结合历史剧情和上下文，以流畅的风格将下面的文本从日文翻译成简体中文：{line}"
-                });
-            }
+                Role = "user",
+                Content = $"参考以下术语表（可为空，格式为src->dst #备注）：{(string.IsNullOrEmpty(dictStr) ? string.Empty : "\n" + dictStr)}\n\n" +
+                            $"根据上述术语表的对应关系和备注，结合历史剧情和上下文，以流畅的风格将下面的文本从日文翻译成简体中文：{line}"
+            });
 
-            messagesStr = SerializePromptMessages(messages);
+            var messagesStr = SerializePromptMessages(messages);
             //Console.WriteLine($"提交的prompt: {messagesStr}");
 
             return $"{{" +

diff --git a/SakuraTranslator/SakuraUtil.cs b/SakuraTranslator/SakuraUtil.cs
@@ -0,0 +1,19 @@
+namespace SakuraTranslate
+{
+    public static class SakuraUtil
+    {
+        public static string FixTranslationEnd(string original, string translation)
+        {
+            if (translation.EndsWith("。") && !original.EndsWith("。"))
+            {
+                translation = translation.Substring(0, translation.Length - "。".Length);
+            }
+            if (translation.EndsWith("。」") && !original.EndsWith("。」"))
+            {
+                translation = translation.Substring(0, translation.Length - "。」".Length) + "」";
+            }
+
+            return translation;
+        }
+    }
+}
diff --git a/SakuraTranslator/TranslationModel.cs b/SakuraTranslator/TranslationModel.cs
@@ -47,5 +47,10 @@ private static TranslationModel GetTranslationModel(string modelName, string mod
                     return TranslationModel.SakuraV1_0;
             }
         }
+
+        private static bool IsOpenAIEndpoint(TranslationModel model)
+        {
+            return !(model == TranslationModel.SakuraV0_8 || model == TranslationModel.SakuraV0_9);
+        }
     }
 }