prompt tuning

Author: cgsv

lib/openai/prompt.ts

@@ -1,9 +1,11 @@
 import { getRandomInt } from "../../utils/fp";
 import {Node} from "../srt"
 import { OpenAIStreamPayload } from "./OpenAIResult";
+import { DEFAULT_PROMPT } from "../../utils/constants";
 
 // gpt返回格式为： '1\n我仍然在问自己，当我离开这个漂浮的城市时，我是否做了正确的事情。\n\n2\n我不仅指工作。'
 export function parse_gpt_resp(content: string, res_keys: number[]) {
+    console.log(content);
     const lines = content.split("\n").map(line => line.trim()).filter(line => line.length > 0);
     const positions = res_keys.map(i => lines.indexOf(String(i)));
     const res: string[] = [];
@@ -12,8 +14,7 @@ export function parse_gpt_resp(content: string, res_keys: number[]) {
         if (p1 === -1) {
             res.push("");
         } else {
-            let next_pos: number | undefined = positions[i+1];
-            if (next_pos === -1) next_pos = undefined;
+            let next_pos = positions.slice(i+1).find(p2 => p2 !== -1);
             res.push(lines.slice(p1+1, next_pos).join('\n'));
         }
     }
@@ -28,13 +29,11 @@ export function nodesToQueryText(nodes: Node[]) {
 
 // 中文， 英文
 function systemMessage(targetLang: string, srcLang?: string, promptTemplate?: string) {
-    if (!srcLang) {
-        if (promptTemplate) return promptTemplate.replace("{{target_lang}}", targetLang);
-        return `你是一个专业的翻译。请逐行翻译上面的文本到${targetLang}，注意保留行号及换行符。`;
-    } else {
-        if (promptTemplate) return promptTemplate.replace("{{target_lang}}", targetLang).replace("{{src_lang}}", srcLang);
-        return `你是一个专业的翻译。请逐行把上面的文本从${srcLang}翻译到${targetLang}，注意保留行号及换行符。`;
+    let prompt = (promptTemplate || DEFAULT_PROMPT).replace("{{target_lang}}", targetLang);
+    if (srcLang) {
+        prompt = prompt.replace("{{src_lang}}", srcLang);
     }
+    return prompt;
 }
 
 const rand4 = () => getRandomInt(1000, 9999);
@@ -59,8 +58,8 @@ export function getPayload(sentences: string[], targetLang: string, srcLang?: st
     const payload: OpenAIStreamPayload = {
         model: "gpt-3.5-turbo",
         messages: [
-            // {role: "system" as const, content: systemMessage(targetLang, srcLang) },
-            {role: "user" as const, content: nodesToQueryText(nodes) + "\n" + systemMessage(targetLang, srcLang, promptTemplate)}],
+            {role: "system" as const, content: systemMessage(targetLang, srcLang, promptTemplate) },
+            {role: "user" as const, content: nodesToQueryText(nodes)}],
         temperature: 0,    // translate task temperature 0?
         top_p: 1,
         frequency_penalty: 0,

middleware.ts

@@ -15,8 +15,8 @@ export async function middleware(req: NextRequest, context: NextFetchEvent) {
   rkey = "tranres_" + await digestMessage(rkey);
   const cached = await redis.get<string[]>(rkey);
 
-  //if (!isDev && cached) {
-  if (cached) {
+  if (!isDev && cached) {
+  //if (cached) {
     console.log("Using cached response " + rkey);
     return NextResponse.json(cached);
   }

pages/srt.tsx

@@ -39,7 +39,7 @@ function curPageNodes(nodes: Node[], curPage: number) {
 
 const sleep = (ms: number) => new Promise((r) => setTimeout(r, ms));
 
-async function traslate_all(nodes: Node[], lang: string, apiKey?: string, notifyResult?: any, useGoogle?: boolean) {
+async function traslate_all(nodes: Node[], lang: string, apiKey?: string, notifyResult?: any, useGoogle?: boolean, promptTemplate?: string) {
     const batches: Node[][] = [];
     for (let i = 0; i < nodes.length; i += PAGE_SIZE) {
         batches.push(nodes.slice(i, i + PAGE_SIZE));
@@ -51,7 +51,7 @@ async function traslate_all(nodes: Node[], lang: string, apiKey?: string, notify
         let success = false;
         for (let i = 0; i < MAX_RETRY && !success; i++) {
             try {
-                const r = await translate_one_batch(batch, lang, apiKey, useGoogle);
+                const r = await translate_one_batch(batch, lang, apiKey, useGoogle, promptTemplate);
                 results.push(...r);
                 success = true;
                 if (notifyResult) {
@@ -72,7 +72,7 @@ async function traslate_all(nodes: Node[], lang: string, apiKey?: string, notify
     return results;
 }
 
-async function translate_one_batch(nodes: Node[], lang: string, apiKey?: string, useGoogle?: boolean) {
+async function translate_one_batch(nodes: Node[], lang: string, apiKey?: string, useGoogle?: boolean, promptTemplate?: string) {
     const sentences = nodes.map(node => node.content);
     // if last sentence ends with ",", remove it
     const lastSentence = sentences[sentences.length - 1];
@@ -88,7 +88,8 @@ async function translate_one_batch(nodes: Node[], lang: string, apiKey?: string,
         body: JSON.stringify({
             "targetLang": lang,
             "sentences": sentences,
-            "apiKey": apiKey
+            "apiKey": apiKey,
+            "promptTemplate": promptTemplate
         })
     };
 
@@ -157,6 +158,11 @@ export default function Srt() {
         else return userLicenceKey;
     }
 
+    const getUserPrompt = () => {
+        const res =  localStorage.getItem("user-prompt-template");
+        if (res) return res;
+    }
+
     const getLang = () => {
         return (document.getElementById("langSelect") as HTMLSelectElement).value;
     }
@@ -227,7 +233,7 @@ export default function Srt() {
     const translateFile = async () => {
         setTransFileStatus({isTranslating: true, transCount: 0});
         try {
-            const newnodes = await traslate_all(nodes, getLang(), getUserKey(), on_trans_result, getUseGoogle());
+            const newnodes = await traslate_all(nodes, getLang(), getUserKey(), on_trans_result, getUseGoogle(), getUserPrompt());
             //download("output.srt", nodesToSrtText(newnodes));
             toast.success(t("translate file successfully"));
         } catch (e) {
@@ -239,7 +245,7 @@ export default function Srt() {
     const translate = async () => {
         setLoading(true);
         try {
-            const newnodes = await translate_one_batch(curPageNodes(nodes, curPage), getLang(), getUserKey(), getUseGoogle());
+            const newnodes = await translate_one_batch(curPageNodes(nodes, curPage), getLang(), getUserKey(), getUseGoogle(), getUserPrompt());
             setTransNodes(nodes => {
                 const nodesCopy = [...nodes];
                 for (let i = 0; i < PAGE_SIZE; i++) {

utils/constants.ts

@@ -1,3 +1,4 @@
 export const RATE_LIMIT_COUNT = 5;
 export const CHECKOUT_URL = "https://coolapps.lemonsqueezy.com/checkout/buy/f53d875a-4424-4ea3-8398-f3559dfaef98";
 export const ENABLE_SHOP = process.env.NEXT_PUBLIC_ENABLE_SHOP === "true";
+export const DEFAULT_PROMPT = "你是一个专业的翻译。请逐行翻译下面的文本到{{target_lang}}，注意保留数字和换行符，请勿自行创建内容，除了翻译，不要输出任何其他文本。"