chore

update readme
add strategy guides
2023-06-03 20:34:33 +08:00 · 2023-06-03 20:28:21 +08:00 · 2023-06-03 20:18:57 +08:00 · 2023-06-03 19:38:24 +08:00 · 2023-06-03 19:28:37 +08:00 · 2023-06-03 17:36:50 +08:00
38 changed files with 684 additions and 217 deletions
--- a/.gitattributes
+++ b/.gitattributes
@@ -0,0 +1,6 @@
+backend-python/rwkv_pip/** linguist-vendored
+backend-python/wkv_cuda_utils/** linguist-vendored
+backend-python/get-pip.py linguist-vendored
+backend-python/convert_model.py linguist-vendored
+build/** linguist-vendored
+frontend/wailsjs/** linguist-generated
--- a/.gitignore
+++ b/.gitignore
@@ -16,3 +16,6 @@ __pycache__
 /backend-python/wkv_cuda
 *.exe
 *.old
+.DS_Store
+*.log.*
+*.log
--- a/1
+++ b/1
@@ -10,6 +10,7 @@ build-windows:

 build-macos:
 	@echo ---- build for macos
+	wails build -ldflags "-s -w"

 dev:
 	wails dev
--- a/README.md
+++ b/README.md
@@ -39,7 +39,8 @@ English | [简体中文](README_ZH.md)
 - Fully compatible with the OpenAI API, making every ChatGPT client an RWKV client. After starting the model,
  open http://127.0.0.1:8000/docs to view more details.
 - Automatic dependency installation, requiring only a lightweight executable program
- User-friendly chat interaction interface included
+- Configs with 2G to 32G VRAM are included, works well on almost all computers
+- User-friendly chat and completion interaction interface included
 - Easy-to-understand and operate parameter configuration
 - Built-in model conversion tool
 - Built-in download management and remote model inspection
@@ -72,6 +73,7 @@ body.json:
 - [x] CUDA operator int8 acceleration
 - [ ] macOS support
 - [ ] Linux support
+- [ ] Local State Cache DB

 ## Related Repositories:

--- a/README_ZH.md
+++ b/README_ZH.md
@@ -14,7 +14,7 @@ API兼容的接口，这意味着一切ChatGPT客户端都是RWKV客户端。

 [English](README.md) | 简体中文

-[视频演示](https://www.bilibili.com/video/BV1hM4y1v76R) | [疑难解答](https://www.bilibili.com/read/cv23921171) | [预览](#Preview) | [下载][download-url]
+[视频演示](https://www.bilibili.com/video/BV1hM4y1v76R) | [疑难解答](https://www.bilibili.com/read/cv23921171) | [预览](#Preview) | [下载][download-url] | [懒人包](https://pan.baidu.com/s/1wchIUHgne3gncIiLIeKBEQ?pwd=1111)

 [license-image]: http://img.shields.io/badge/license-MIT-blue.svg

@@ -39,7 +39,8 @@ API兼容的接口，这意味着一切ChatGPT客户端都是RWKV客户端。
 - RWKV模型管理，一键启动
 - 与OpenAI API完全兼容，一切ChatGPT客户端，都是RWKV客户端。启动模型后，打开 http://127.0.0.1:8000/docs 查看详细内容
 - 全自动依赖安装，你只需要一个轻巧的可执行程序
- 自带用户友好的聊天交互页面
+- 预设了2G至32G显存的配置，几乎在各种电脑上工作良好
+- 自带用户友好的聊天和补全交互页面
 - 易于理解和操作的参数配置
 - 内置模型转换工具
 - 内置下载管理和远程模型检视
@@ -72,6 +73,7 @@ body.json:
 - [x] CUDA算子int8提速
 - [ ] macOS支持
 - [ ] linux支持
+- [ ] 本地状态缓存数据库

 ## 相关仓库:

--- a/backend-golang/app.go
+++ b/backend-golang/app.go
@@ -5,6 +5,7 @@ import (
 	"net/http"
 	"os"
 	"os/exec"
+	"path/filepath"
 	"runtime"

 	"github.com/minio/selfupdate"
@@ -13,7 +14,9 @@ import (

 // App struct
 type App struct {
-	ctx context.Context
+	ctx       context.Context
+	exDir     string
+	cmdPrefix string
 }

 // NewApp creates a new App application struct
@@ -25,6 +28,14 @@ func NewApp() *App {
 // so we can call the runtime methods
 func (a *App) OnStartup(ctx context.Context) {
 	a.ctx = ctx
+	a.exDir = ""
+	a.cmdPrefix = ""
+
+	if runtime.GOOS == "darwin" {
+		ex, _ := os.Executable()
+		a.exDir = filepath.Dir(ex) + "/../../../"
+		a.cmdPrefix = "cd " + a.exDir + " && "
+	}

 	a.downloadLoop()
 }
--- a/backend-golang/download.go
+++ b/backend-golang/download.go
@@ -9,7 +9,7 @@ import (
 )

 func (a *App) DownloadFile(path string, url string) error {
-	_, err := grab.Get(path, url)
+	_, err := grab.Get(a.exDir+path, url)
 	if err != nil {
 		return err
 	}
@@ -81,7 +81,7 @@ func (a *App) AddToDownloadList(path string, url string) {
 		downloadList = append(downloadList, DownloadStatus{
 			resp:        nil,
 			Name:        filepath.Base(path),
-			Path:        path,
+			Path:        a.exDir + path,
 			Url:         url,
 			Downloading: true,
 		})
--- a/backend-golang/file.go
+++ b/backend-golang/file.go
@@ -2,12 +2,13 @@ package backend_golang

 import (
 	"encoding/json"
-	"fmt"
+	"errors"
 	"io"
 	"os"
 	"os/exec"
 	"path/filepath"
 	"runtime"
+	"strings"
 	"time"
 )

@@ -17,14 +18,14 @@ func (a *App) SaveJson(fileName string, jsonData any) error {
 		return err
 	}

-	if err := os.WriteFile(fileName, text, 0644); err != nil {
+	if err := os.WriteFile(a.exDir+fileName, text, 0644); err != nil {
 		return err
 	}
 	return nil
 }

 func (a *App) ReadJson(fileName string) (any, error) {
-	file, err := os.ReadFile(fileName)
+	file, err := os.ReadFile(a.exDir + fileName)
 	if err != nil {
 		return nil, err
 	}
@@ -39,7 +40,7 @@ func (a *App) ReadJson(fileName string) (any, error) {
 }

 func (a *App) FileExists(fileName string) bool {
-	_, err := os.Stat(fileName)
+	_, err := os.Stat(a.exDir + fileName)
 	return err == nil
 }

@@ -51,7 +52,7 @@ type FileInfo struct {
 }

 func (a *App) ReadFileInfo(fileName string) (FileInfo, error) {
-	info, err := os.Stat(fileName)
+	info, err := os.Stat(a.exDir + fileName)
 	if err != nil {
 		return FileInfo{}, err
 	}
@@ -64,7 +65,7 @@ func (a *App) ReadFileInfo(fileName string) (FileInfo, error) {
 }

 func (a *App) ListDirFiles(dirPath string) ([]FileInfo, error) {
-	files, err := os.ReadDir(dirPath)
+	files, err := os.ReadDir(a.exDir + dirPath)
 	if err != nil {
 		return nil, err
 	}
@@ -86,7 +87,7 @@ func (a *App) ListDirFiles(dirPath string) ([]FileInfo, error) {
 }

 func (a *App) DeleteFile(path string) error {
-	err := os.Remove(path)
+	err := os.Remove(a.exDir + path)
 	if err != nil {
 		return err
 	}
@@ -94,13 +95,18 @@ func (a *App) DeleteFile(path string) error {
 }

 func (a *App) CopyFile(src string, dst string) error {
-	sourceFile, err := os.Open(src)
+	sourceFile, err := os.Open(a.exDir + src)
 	if err != nil {
 		return err
 	}
 	defer sourceFile.Close()

-	destFile, err := os.Create(dst)
+	err = os.MkdirAll(a.exDir+dst[:strings.LastIndex(dst, "/")], 0755)
+	if err != nil {
+		return err
+	}
+
+	destFile, err := os.Create(a.exDir + dst)
 	if err != nil {
 		return err
 	}
@@ -114,7 +120,7 @@ func (a *App) CopyFile(src string, dst string) error {
 }

 func (a *App) OpenFileFolder(path string) error {
-	absPath, err := filepath.Abs(path)
+	absPath, err := filepath.Abs(a.exDir + path)
 	if err != nil {
 		return err
 	}
@@ -125,10 +131,16 @@ func (a *App) OpenFileFolder(path string) error {
 		if err != nil {
 			return err
 		}
+		return nil
 	case "darwin":
-		fmt.Println("Running on macOS")
+		cmd := exec.Command("open", "-R", absPath)
+		err := cmd.Run()
+		if err != nil {
+			return err
+		}
+		return nil
 	case "linux":
-		fmt.Println("Running on Linux")
+		println("unsupported OS")
 	}
-	return nil
+	return errors.New("unsupported OS")
 }
--- a/backend-golang/rwkv.go
+++ b/backend-golang/rwkv.go
@@ -7,36 +7,48 @@ import (
 	"strconv"
 )

-func (a *App) StartServer(port int, host string) (string, error) {
-	python, err := GetPython()
+func (a *App) StartServer(python string, port int, host string) (string, error) {
+	var err error
+	if python == "" {
+		python, err = GetPython()
+	}
 	if err != nil {
 		return "", err
 	}
 	return Cmd(python, "./backend-python/main.py", strconv.Itoa(port), host)
 }

-func (a *App) ConvertModel(modelPath string, strategy string, outPath string) (string, error) {
-	python, err := GetPython()
+func (a *App) ConvertModel(python string, modelPath string, strategy string, outPath string) (string, error) {
+	var err error
+	if python == "" {
+		python, err = GetPython()
+	}
 	if err != nil {
 		return "", err
 	}
 	return Cmd(python, "./backend-python/convert_model.py", "--in", modelPath, "--out", outPath, "--strategy", strategy)
 }

-func (a *App) DepCheck() error {
-	python, err := GetPython()
+func (a *App) DepCheck(python string) error {
+	var err error
+	if python == "" {
+		python, err = GetPython()
+	}
 	if err != nil {
 		return err
 	}
-	out, err := exec.Command(python, "./backend-python/dep_check.py").CombinedOutput()
+	out, err := exec.Command(python, a.exDir+"./backend-python/dep_check.py").CombinedOutput()
 	if err != nil {
 		return errors.New("DepCheck Error: " + string(out))
 	}
 	return nil
 }

-func (a *App) InstallPyDep(cnMirror bool) (string, error) {
-	python, err := GetPython()
+func (a *App) InstallPyDep(python string, cnMirror bool) (string, error) {
+	var err error
+	if python == "" {
+		python, err = GetPython()
+	}
 	if err != nil {
 		return "", err
 	}
@@ -51,7 +63,11 @@ func (a *App) InstallPyDep(cnMirror bool) (string, error) {
 	if err != nil {
 		return "", err
 	}
-	_, err = Cmd(python, "-m", "pip", "install", "torch==1.13.1", "torchvision==0.14.1", "torchaudio==0.13.1", "--index-url", "https://download.pytorch.org/whl/cu117")
+	if runtime.GOOS == "windows" {
+		_, err = Cmd(python, "-m", "pip", "install", "torch==1.13.1", "torchvision==0.14.1", "torchaudio==0.13.1", "--index-url", "https://download.pytorch.org/whl/cu117")
+	} else {
+		_, err = Cmd(python, "-m", "pip", "install", "torch", "torchvision", "torchaudio")
+	}
 	if err != nil {
 		return "", err
 	}
--- a/backend-golang/utils.go
+++ b/backend-golang/utils.go
@@ -15,7 +15,8 @@ import (
 )

 func Cmd(args ...string) (string, error) {
-	if runtime.GOOS == "windows" {
+	switch platform := runtime.GOOS; platform {
+	case "windows":
 		_, err := os.Stat("cmd-helper.bat")
 		if err != nil {
 			if err := os.WriteFile("./cmd-helper.bat", []byte("start %*"), 0644); err != nil {
@@ -26,13 +27,33 @@ func Cmd(args ...string) (string, error) {
 		if err != nil {
 			return "", err
 		}
+
+		for _, arg := range args {
+			if strings.Contains(arg, " ") && strings.Contains(cmdHelper, " ") {
+				return "", errors.New("path contains space") // golang bug https://github.com/golang/go/issues/17149#issuecomment-473976818
+			}
+		}
+
 		cmd := exec.Command(cmdHelper, args...)
 		out, err := cmd.CombinedOutput()
 		if err != nil {
 			return "", err
 		}
 		return string(out), nil
-	} else {
+	case "darwin":
+		ex, err := os.Executable()
+		if err != nil {
+			return "", err
+		}
+		exDir := filepath.Dir(ex) + "/../../../"
+		cmd := exec.Command("osascript", "-e", `tell application 'Terminal' to do script '`+"cd "+exDir+" && "+strings.Join(args, " ")+`'`)
+		err = cmd.Start()
+		if err != nil {
+			return "", err
+		}
+		cmd.Wait()
+		return "", nil
+	case "linux":
 		cmd := exec.Command(args[0], args[1:]...)
 		err := cmd.Start()
 		if err != nil {
@@ -41,9 +62,19 @@ func Cmd(args ...string) (string, error) {
 		cmd.Wait()
 		return "", nil
 	}
+	return "", errors.New("unsupported OS")
 }

 func CopyEmbed(efs embed.FS) error {
+	prefix := ""
+	if runtime.GOOS == "darwin" {
+		ex, err := os.Executable()
+		if err != nil {
+			return err
+		}
+		prefix = filepath.Dir(ex) + "/../../../"
+	}
+
 	err := fs.WalkDir(efs, ".", func(path string, d fs.DirEntry, err error) error {
 		if d.IsDir() {
 			return nil
@@ -56,6 +87,7 @@ func CopyEmbed(efs embed.FS) error {
 			return err
 		}

+		path = prefix + path
 		err = os.MkdirAll(path[:strings.LastIndex(path, "/")], 0755)
 		if err != nil {
 			return err
--- a/backend-python/dep_check.py
+++ b/backend-python/dep_check.py
@@ -1,8 +1,6 @@
-import cyac
 import GPUtil
 import torch
 import rwkv
-import langchain
 import fastapi
 import uvicorn
 import sse_starlette
--- a/backend-python/main.py
+++ b/backend-python/main.py
@@ -4,17 +4,18 @@ import sys
 sys.path.append(os.path.dirname(os.path.realpath(__file__)))

 import psutil
-from fastapi import FastAPI
+from fastapi import Depends, FastAPI
 from fastapi.middleware.cors import CORSMiddleware
 import uvicorn

 from utils.rwkv import *
 from utils.torch import *
 from utils.ngrok import *
+from utils.log import log_middleware
 from routes import completion, config, state_cache
 import global_var

-app = FastAPI()
+app = FastAPI(dependencies=[Depends(log_middleware)])

 app.add_middleware(
    CORSMiddleware,
@@ -42,7 +43,7 @@ def init():

@app.get("/")
 def read_root():
-    return {"Hello": "World!", "pid": os.getpid()}
+    return {"Hello": "World!"}


@app.post("/exit")
@@ -60,7 +61,7 @@ def debug():
        strategy="cuda fp16",
        tokens_path="20B_tokenizer.json",
    )
-    d = model.tokenizer.decode([])
+    d = model.pipeline.decode([])
    print(d)


--- a/backend-python/routes/completion.py
+++ b/backend-python/routes/completion.py
@@ -7,6 +7,7 @@ from fastapi import APIRouter, Request, status, HTTPException
 from sse_starlette.sse import EventSourceResponse
 from pydantic import BaseModel
 from utils.rwkv import *
+from utils.log import quick_log
 import global_var

 router = APIRouter()
@@ -26,6 +27,8 @@ class ChatCompletionBody(ModelConfigBody):

 completion_lock = Lock()

+requests_num = 0
+

@router.post("/v1/chat/completions")
@router.post("/chat/completions")
@@ -59,28 +62,28 @@ The following is a coherent verbose detailed conversation between a girl named {
 {bot} usually gives {user} kind, helpful and informative advices.\n
 """
        if user == "Bob"
-        else ""
+        else f"{user}{interface} hi\n\n{bot}{interface} Hi. I am your assistant and I will provide expert full response in full details. Please feel free to ask any question and I will always answer it.\n\n"
    )
    for message in body.messages:
        if message.role == "system":
            completion_text = (
                f"The following is a coherent verbose detailed conversation between a girl named {bot} and her friend {user}. "
                if user == "Bob"
-                else ""
+                else f"{user}{interface} hi\n\n{bot}{interface} Hi. "
                + message.content.replace("\\n", "\n")
                .replace("\r\n", "\n")
                .replace("\n\n", "\n")
                .replace("\n", " ")
                .strip()
-                .replace("You are", f"{bot} is")
-                .replace("you are", f"{bot} is")
-                .replace("You're", f"{bot} is")
-                .replace("you're", f"{bot} is")
-                .replace("You", f"{bot}")
-                .replace("you", f"{bot}")
-                .replace("Your", f"{bot}'s")
-                .replace("your", f"{bot}'s")
-                .replace("你", f"{bot}")
+                .replace("You are", f"{bot} is" if user == "Bob" else "I am")
+                .replace("you are", f"{bot} is" if user == "Bob" else "I am")
+                .replace("You're", f"{bot} is" if user == "Bob" else "I'm")
+                .replace("you're", f"{bot} is" if user == "Bob" else "I'm")
+                .replace("You", f"{bot}" if user == "Bob" else "I")
+                .replace("you", f"{bot}" if user == "Bob" else "I")
+                .replace("Your", f"{bot}'s" if user == "Bob" else "My")
+                .replace("your", f"{bot}'s" if user == "Bob" else "my")
+                .replace("你", f"{bot}" if user == "Bob" else "我")
                + "\n\n"
            )
            break
@@ -106,12 +109,32 @@ The following is a coherent verbose detailed conversation between a girl named {
    completion_text += f"{bot}{interface}"

    async def eval_rwkv():
+        global requests_num
+        requests_num = requests_num + 1
+        quick_log(request, None, "Start Waiting. RequestsNum: " + str(requests_num))
        while completion_lock.locked():
            if await request.is_disconnected():
+                requests_num = requests_num - 1
+                print(f"{request.client} Stop Waiting (Lock)")
+                quick_log(
+                    request,
+                    None,
+                    "Stop Waiting (Lock). RequestsNum: " + str(requests_num),
+                )
                return
            await asyncio.sleep(0.1)
        else:
            completion_lock.acquire()
+            if await request.is_disconnected():
+                completion_lock.release()
+                requests_num = requests_num - 1
+                print(f"{request.client} Stop Waiting (Lock)")
+                quick_log(
+                    request,
+                    None,
+                    "Stop Waiting (Lock). RequestsNum: " + str(requests_num),
+                )
+                return
            set_rwkv_config(model, global_var.get(global_var.Model_Config))
            set_rwkv_config(model, body)
            if body.stream:
@@ -135,9 +158,21 @@ The following is a coherent verbose detailed conversation between a girl named {
                        }
                    )
                # torch_gc()
+                requests_num = requests_num - 1
                completion_lock.release()
                if await request.is_disconnected():
+                    print(f"{request.client} Stop Waiting")
+                    quick_log(
+                        request,
+                        body,
+                        response + "\nStop Waiting. RequestsNum: " + str(requests_num),
+                    )
                    return
+                quick_log(
+                    request,
+                    body,
+                    response + "\nFinished. RequestsNum: " + str(requests_num),
+                )
                yield json.dumps(
                    {
                        "response": response,
@@ -161,9 +196,21 @@ The following is a coherent verbose detailed conversation between a girl named {
                    if await request.is_disconnected():
                        break
                # torch_gc()
+                requests_num = requests_num - 1
                completion_lock.release()
                if await request.is_disconnected():
+                    print(f"{request.client} Stop Waiting")
+                    quick_log(
+                        request,
+                        body,
+                        response + "\nStop Waiting. RequestsNum: " + str(requests_num),
+                    )
                    return
+                quick_log(
+                    request,
+                    body,
+                    response + "\nFinished. RequestsNum: " + str(requests_num),
+                )
                yield {
                    "response": response,
                    "model": "rwkv",
@@ -182,7 +229,10 @@ The following is a coherent verbose detailed conversation between a girl named {
    if body.stream:
        return EventSourceResponse(eval_rwkv())
    else:
-        return await eval_rwkv().__anext__()
+        try:
+            return await eval_rwkv().__anext__()
+        except StopAsyncIteration:
+            return None


 class CompletionBody(ModelConfigBody):
@@ -203,12 +253,32 @@ async def completions(body: CompletionBody, request: Request):
        raise HTTPException(status.HTTP_400_BAD_REQUEST, "prompt not found")

    async def eval_rwkv():
+        global requests_num
+        requests_num = requests_num + 1
+        quick_log(request, None, "Start Waiting. RequestsNum: " + str(requests_num))
        while completion_lock.locked():
            if await request.is_disconnected():
+                requests_num = requests_num - 1
+                print(f"{request.client} Stop Waiting (Lock)")
+                quick_log(
+                    request,
+                    None,
+                    "Stop Waiting (Lock). RequestsNum: " + str(requests_num),
+                )
                return
            await asyncio.sleep(0.1)
        else:
            completion_lock.acquire()
+            if await request.is_disconnected():
+                completion_lock.release()
+                requests_num = requests_num - 1
+                print(f"{request.client} Stop Waiting (Lock)")
+                quick_log(
+                    request,
+                    None,
+                    "Stop Waiting (Lock). RequestsNum: " + str(requests_num),
+                )
+                return
            set_rwkv_config(model, global_var.get(global_var.Model_Config))
            set_rwkv_config(model, body)
            if body.stream:
@@ -229,9 +299,21 @@ async def completions(body: CompletionBody, request: Request):
                        }
                    )
                # torch_gc()
+                requests_num = requests_num - 1
                completion_lock.release()
                if await request.is_disconnected():
+                    print(f"{request.client} Stop Waiting")
+                    quick_log(
+                        request,
+                        body,
+                        response + "\nStop Waiting. RequestsNum: " + str(requests_num),
+                    )
                    return
+                quick_log(
+                    request,
+                    body,
+                    response + "\nFinished. RequestsNum: " + str(requests_num),
+                )
                yield json.dumps(
                    {
                        "response": response,
@@ -252,9 +334,21 @@ async def completions(body: CompletionBody, request: Request):
                    if await request.is_disconnected():
                        break
                # torch_gc()
+                requests_num = requests_num - 1
                completion_lock.release()
                if await request.is_disconnected():
+                    print(f"{request.client} Stop Waiting")
+                    quick_log(
+                        request,
+                        body,
+                        response + "\nStop Waiting. RequestsNum: " + str(requests_num),
+                    )
                    return
+                quick_log(
+                    request,
+                    body,
+                    response + "\nFinished. RequestsNum: " + str(requests_num),
+                )
                yield {
                    "response": response,
                    "model": "rwkv",
@@ -270,4 +364,7 @@ async def completions(body: CompletionBody, request: Request):
    if body.stream:
        return EventSourceResponse(eval_rwkv())
    else:
-        return await eval_rwkv().__anext__()
+        try:
+            return await eval_rwkv().__anext__()
+        except StopAsyncIteration:
+            return None
--- a/backend-python/routes/config.py
+++ b/backend-python/routes/config.py
@@ -1,8 +1,7 @@
 import pathlib

-from fastapi import APIRouter, HTTPException, Response, status
+from fastapi import APIRouter, HTTPException, Response, status as Status
 from pydantic import BaseModel
-from langchain.llms import RWKV
 from utils.rwkv import *
 from utils.torch import *
 import global_var
@@ -33,7 +32,7 @@ class SwitchModelBody(BaseModel):
@router.post("/switch-model")
 def switch_model(body: SwitchModelBody, response: Response):
    if global_var.get(global_var.Model_Status) is global_var.ModelStatus.Loading:
-        response.status_code = status.HTTP_304_NOT_MODIFIED
+        response.status_code = Status.HTTP_304_NOT_MODIFIED
        return

    global_var.set(global_var.Model_Status, global_var.ModelStatus.Offline)
@@ -55,7 +54,7 @@ def switch_model(body: SwitchModelBody, response: Response):
    except Exception as e:
        print(e)
        global_var.set(global_var.Model_Status, global_var.ModelStatus.Offline)
-        raise HTTPException(status.HTTP_500_INTERNAL_SERVER_ERROR, "failed to load")
+        raise HTTPException(Status.HTTP_500_INTERNAL_SERVER_ERROR, "failed to load")

    if global_var.get(global_var.Model_Config) is None:
        global_var.set(
--- a/backend-python/routes/state_cache.py
+++ b/backend-python/routes/state_cache.py
@@ -3,6 +3,7 @@ from fastapi import APIRouter, HTTPException, Response, status
 from pydantic import BaseModel
 import gc
 import copy
+import torch

 router = APIRouter()

@@ -41,10 +42,14 @@ def add_state(body: AddStateBody):
        raise HTTPException(status.HTTP_400_BAD_REQUEST, "trie not loaded")

    id = trie.insert(body.prompt)
+    device = body.state[0].device
    dtrie[id] = {
        "tokens": copy.deepcopy(body.tokens),
-        "state": copy.deepcopy(body.state),
+        "state": [tensor.cpu() for tensor in body.state]
+        if device != torch.device("cpu")
+        else copy.deepcopy(body.state),
        "logits": copy.deepcopy(body.logits),
+        "device": device,
    }

    return "success"
@@ -77,11 +82,15 @@ def longest_prefix_state(body: LongestPrefixStateBody):
        pass
    if id != -1:
        v = dtrie[id]
+        device = v["device"]
        return {
            "prompt": trie[id],
            "tokens": v["tokens"],
-            "state": v["state"],
+            "state": [tensor.to(device) for tensor in v["state"]]
+            if device != torch.device("cpu")
+            else v["state"],
            "logits": v["logits"],
+            "device": device,
        }
    else:
        return {"prompt": "", "tokens": [], "state": None, "logits": None}
--- a/backend-python/utils/log.py
+++ b/backend-python/utils/log.py
@@ -0,0 +1,32 @@
+import json
+import logging
+from typing import Any
+from fastapi import Request
+
+
+logger = logging.getLogger()
+logger.setLevel(logging.INFO)
+formatter = logging.Formatter("%(asctime)s - %(levelname)s\n%(message)s")
+fh = logging.handlers.RotatingFileHandler(
+    "api.log", mode="a", maxBytes=3 * 1024 * 1024, backupCount=3
+)
+fh.setFormatter(formatter)
+logger.addHandler(fh)
+
+
+def quick_log(request: Request, body: Any, response: str):
+    logger.info(
+        f"Client: {request.client}\nUrl: {request.url}\n"
+        + (
+            f"Body: {json.dumps(body.__dict__, default=vars, ensure_ascii=False)}\n"
+            if body
+            else ""
+        )
+        + (f"Response:\n{response}\n" if response else "")
+    )
+
+
+async def log_middleware(request: Request):
+    logger.info(
+        f"Client: {request.client}\nUrl: {request.url}\nBody: {await request.body()}\n"
+    )
--- a/backend-python/utils/rwkv.py
+++ b/backend-python/utils/rwkv.py
@@ -3,13 +3,14 @@ import pathlib
 import copy
 from typing import Dict, List
 from fastapi import HTTPException
-from pydantic import BaseModel
+from pydantic import BaseModel, Field
 from rwkv_pip.utils import PIPELINE
 from routes import state_cache


 END_OF_TEXT = 0
 END_OF_LINE = 187
+END_OF_LINE_DOUBLE = 535


 os.environ["TORCH_EXTENSIONS_DIR"] = f"{pathlib.Path(__file__).parent.parent.resolve()}"
@@ -50,28 +51,38 @@ class RWKV:
        self.preload()

    def preload(self):
-        if self.user == "Bob":
-            bot = self.bot
-            user = self.user
-            preset_system = f"""
+        interface = self.interface
+        user = self.user
+        bot = self.bot
+        preset_system = (
+            f"""
 The following is a coherent verbose detailed conversation between a girl named {bot} and her friend {user}. \
 {bot} is very intelligent, creative and friendly. \
 {bot} is unlikely to disagree with {user}, and {bot} doesn't like to ask {user} questions. \
 {bot} likes to tell {user} a lot about herself and her opinions. \
 {bot} usually gives {user} kind, helpful and informative advices.\n
 """
-            logits = self.run_rnn(self.pipeline.encode(preset_system))
-            try:
-                state_cache.add_state(
-                    state_cache.AddStateBody(
-                        prompt=preset_system,
-                        tokens=self.model_tokens,
-                        state=self.model_state,
-                        logits=logits,
-                    )
+            if self.user == "Bob"
+            else f"{user}{interface} hi\n\n{bot}{interface} Hi. I am your assistant and I will provide expert full response in full details. Please feel free to ask any question and I will always answer it.\n\n"
+        )
+        logits = self.run_rnn(self.fix_tokens(self.pipeline.encode(preset_system)))
+        try:
+            state_cache.add_state(
+                state_cache.AddStateBody(
+                    prompt=preset_system,
+                    tokens=self.model_tokens,
+                    state=self.model_state,
+                    logits=logits,
                )
-            except HTTPException:
-                pass
+            )
+        except HTTPException:
+            pass
+
+    # Model only saw '\n\n' as [187, 187] before, but the tokenizer outputs [535] for it at the end
+    def fix_tokens(self, tokens):
+        if len(tokens) > 0 and tokens[-1] == END_OF_LINE_DOUBLE:
+            tokens = tokens[:-1] + [END_OF_LINE, END_OF_LINE]
+        return tokens

    def run_rnn(self, _tokens: List[str], newline_adj: int = 0):
        tokens = [int(x) for x in _tokens]
@@ -108,7 +119,7 @@ The following is a coherent verbose detailed conversation between a girl named {
            logits = copy.deepcopy(cache["logits"])

        if delta_prompt != "":
-            logits = self.run_rnn(self.pipeline.encode(delta_prompt))
+            logits = self.run_rnn(self.fix_tokens(self.pipeline.encode(delta_prompt)))
            try:
                state_cache.add_state(
                    state_cache.AddStateBody(
@@ -182,23 +193,23 @@ The following is a coherent verbose detailed conversation between a girl named {


 class ModelConfigBody(BaseModel):
-    max_tokens: int = None
-    temperature: float = None
-    top_p: float = None
-    presence_penalty: float = None
-    frequency_penalty: float = None
+    max_tokens: int = Field(default=None, gt=0, le=102400)
+    temperature: float = Field(default=None, ge=0, le=2)
+    top_p: float = Field(default=None, ge=0, le=1)
+    presence_penalty: float = Field(default=None, ge=-2, le=2)
+    frequency_penalty: float = Field(default=None, ge=-2, le=2)


 def set_rwkv_config(model: RWKV, body: ModelConfigBody):
-    if body.max_tokens:
+    if body.max_tokens is not None:
        model.max_tokens_per_generation = body.max_tokens
-    if body.temperature:
+    if body.temperature is not None:
        model.temperature = body.temperature
-    if body.top_p:
+    if body.top_p is not None:
        model.top_p = body.top_p
-    if body.presence_penalty:
+    if body.presence_penalty is not None:
        model.penalty_alpha_presence = body.presence_penalty
-    if body.frequency_penalty:
+    if body.frequency_penalty is not None:
        model.penalty_alpha_frequency = body.frequency_penalty


--- a/frontend/src/_locales/zh-hans/main.json
+++ b/frontend/src/_locales/zh-hans/main.json
@@ -114,6 +114,7 @@
  "Catgirl": "猫娘",
  "Explain Code": "代码解释",
  "Werewolf": "狼人杀",
+  "Instruction": "指令",
  "Blank": "空白",
  "The following is an epic science fiction masterpiece that is immortalized, with delicate descriptions and grand depictions of interstellar civilization wars.\nChapter 1.\n": "以下是不朽的科幻史诗巨著，描写细腻，刻画了宏大的星际文明战争。\n第一章\n",
  "The following is a conversation between a cat girl and her owner. The cat girl is a humanized creature that behaves like a cat but is humanoid. At the end of each sentence in the dialogue, she will add \"Meow~\". In the following content, Bob represents the owner and Alice represents the cat girl.\n\nBob: Hello.\n\nAlice: I'm here, meow~.\n\nBob: Can you tell jokes?": "以下是一位猫娘的主人和猫娘的对话内容，猫娘是一种拟人化的生物，其行为似猫但类人，在每一句对话末尾都会加上\"喵~\"。以下内容中，Bob代表主人，Alice代表猫娘。\n\nBob: 你好\n\nAlice: 主人我在哦，喵~\n\nBob: 你会讲笑话吗?",
@@ -124,5 +125,18 @@
  "There is currently a game of Werewolf with six players, including a Seer (who can check identities at night), two Werewolves (who can choose someone to kill at night), a Bodyguard (who can choose someone to protect at night), two Villagers (with no special abilities), and a game host. Bob will play as Player 1, Alice will play as Players 2-6 and the game host, and they will begin playing together. Every night, the host will ask Bob for his action and simulate the actions of the other players. During the day, the host will oversee the voting process and ask Bob for his vote. \n\nAlice: Next, I will act as the game host and assign everyone their roles, including randomly assigning yours. Then, I will simulate the actions of Players 2-6 and let you know what happens each day. Based on your assigned role, you can tell me your actions and I will let you know the corresponding results each day.\n\nBob: Okay, I understand. Let's begin. Please assign me a role. Am I the Seer, Werewolf, Villager, or Bodyguard?\n\nAlice: You are the Seer. Now that night has fallen, please choose a player to check his identity.\n\nBob: Tonight, I want to check Player 2 and find out his role.": "现在有一场六人狼人杀游戏，包括一名预言家(可以在夜晚查验身份)，两名狼人(可以在夜晚选择杀人)，一名守卫(可以在夜晚选择要守护的人)，两名平民（无技能)，一名主持人，以下内容中Bob将扮演其中的1号玩家，Alice来扮演2-6号玩家，以及主持人，并开始与Bob进行游戏，主持人每晚都会询问Bob的行动，并模拟其他人的行动，在白天则要主持投票，并同样询问Bob投票对象，公布投票结果。\n\nAlice: 接下来，我将首先作为主持人进行角色分配，并给你赋予随机的角色，之后我将模拟2-6号玩家进行行动，告知你每天的动态，根据你被分配的角色，你可以回复我你做的行动，我会告诉你每天对应的结果\n\nBob: 好的，我明白了，那么开始吧。请先给我一个角色身份。我是预言家，狼人，平民，守卫中的哪一个呢？\n\nAlice: 你的身份是预言家。现在夜晚降临，请选择你要查验的玩家。\n\nBob: 今晚我要验2号玩家，他是什么身份？",
  "Writer, Translator, Role-playing": "写作，翻译，角色扮演",
  "Chinese Kongfu": "情境冒险",
-  "Allow external access to the API (service must be restarted)": "允许外部访问API (必须重启服务)"
+  "Allow external access to the API (service must be restarted)": "允许外部访问API (必须重启服务)",
+  "Custom": "自定义",
+  "Reset All Configs": "重置所有配置",
+  "Cancel": "取消",
+  "Confirm": "确认",
+  "Are you sure you want to reset all configs? This will obtain the latest preset configs, but will override your custom configs and cannot be undone.": "你确定要重置所有配置吗？这会获取最新的预设配置，但会覆盖你的自定义配置，并且无法撤销",
+  "Advanced": "高级",
+  "Custom Python Path": "自定义Python路径",
+  "Custom Models Path": "自定义模型路径",
+  "MacOS is not supported yet, please convert manually.": "暂不支持MacOS, 请手动转换",
+  "Microsoft Visual C++ Redistributable is not installed, would you like to download it?": "微软VC++组件未安装, 是否下载?",
+  "Path Cannot Contain Space": "路径不能包含空格",
+  "Failed to switch model, please try starting the program with administrator privileges.": "切换模型失败, 请尝试以管理员权限启动程序",
+  "Current Strategy": "当前Strategy"
 }
--- a/frontend/src/assets/images/logo.jpg
+++ b/frontend/src/assets/images/logo.jpg
--- a/frontend/src/assets/images/strategy.jpg
+++ b/frontend/src/assets/images/strategy.jpg
--- a/frontend/src/assets/images/strategy_zh.jpg
+++ b/frontend/src/assets/images/strategy_zh.jpg
--- a/frontend/src/components/Labeled.tsx
+++ b/frontend/src/components/Labeled.tsx
@@ -5,17 +5,23 @@ import classnames from 'classnames';
 export const Labeled: FC<{
  label: string;
  desc?: string | null,
+  descComponent?: ReactElement,
  content: ReactElement,
  flex?: boolean,
  spaceBetween?: boolean,
-  breakline?: boolean
+  breakline?: boolean,
+  onMouseEnter?: () => void
+  onMouseLeave?: () => void
 }> = ({
  label,
  desc,
+  descComponent,
  content,
  flex,
  spaceBetween,
-  breakline
+  breakline,
+  onMouseEnter,
+  onMouseLeave
 }) => {
  return (
    <div className={classnames(
@@ -24,11 +30,11 @@ export const Labeled: FC<{
      breakline ? 'flex-col' : '',
      spaceBetween && 'justify-between')
    }>
-      {desc ?
-        <Tooltip content={desc} showDelay={0} hideDelay={0} relationship="description">
-          <Label>{label}</Label>
+      {(desc || descComponent) ?
+        <Tooltip content={descComponent ? descComponent : desc!} showDelay={0} hideDelay={0} relationship="description">
+          <Label onMouseEnter={onMouseEnter} onMouseLeave={onMouseLeave}>{label}</Label>
        </Tooltip> :
-        <Label>{label}</Label>
+        <Label onMouseEnter={onMouseEnter} onMouseLeave={onMouseLeave}>{label}</Label>
      }
      {content}
    </div>
--- a/frontend/src/components/RunButton.tsx
+++ b/frontend/src/components/RunButton.tsx
@@ -12,13 +12,12 @@ import { Button } from '@fluentui/react-components';
 import { observer } from 'mobx-react-lite';
 import { exit, getStatus, readRoot, switchModel, updateConfig } from '../apis';
 import { toast } from 'react-toastify';
-import manifest from '../../../manifest.json';
 import { getStrategy, getSupportedCustomCudaFile, saveCache, toastWithButton } from '../utils';
 import { useTranslation } from 'react-i18next';
 import { ToolTipButton } from './ToolTipButton';
 import { Play16Regular, Stop16Regular } from '@fluentui/react-icons';
 import { useNavigate } from 'react-router';
-import { WindowShow } from '../../wailsjs/runtime/runtime';
+import { BrowserOpenURL, WindowShow } from '../../wailsjs/runtime/runtime';

 const mainButtonText = {
  [ModelStatus.Offline]: 'Run',
@@ -51,7 +50,7 @@ export const RunButton: FC<{ onClickRun?: MouseEventHandler, iconMode?: boolean
      let modelPath = '';
      if (modelConfig && modelConfig.modelParameters) {
        modelName = modelConfig.modelParameters.modelName;
-        modelPath = `./${manifest.localModelDir}/${modelName}`;
+        modelPath = `${commonStore.settings.customModelsPath}/${modelName}`;
      } else {
        toast(t('Model Config Exception'), { type: 'error' });
        commonStore.setStatus({ status: ModelStatus.Offline });
@@ -60,7 +59,7 @@ export const RunButton: FC<{ onClickRun?: MouseEventHandler, iconMode?: boolean

      if (!commonStore.depComplete) {
        let depErrorMsg = '';
-        await DepCheck().catch((e) => {
+        await DepCheck(commonStore.settings.customPythonPath).catch((e) => {
          depErrorMsg = e.message || e;
          WindowShow();
          if (depErrorMsg === 'python zip not found') {
@@ -71,10 +70,16 @@ export const RunButton: FC<{ onClickRun?: MouseEventHandler, iconMode?: boolean
              AddToDownloadList('python-3.10.11-embed-amd64.zip', 'https://www.python.org/ftp/python/3.10.11/python-3.10.11-embed-amd64.zip');
            });
          } else if (depErrorMsg.includes('DepCheck Error')) {
-            toastWithButton(t('Python dependencies are incomplete, would you like to install them?'), t('Install'), () => {
-              InstallPyDep(commonStore.settings.cnMirror);
-              setTimeout(WindowShow, 1000);
-            });
+            if (depErrorMsg.includes('vc_redist')) {
+              toastWithButton(t('Microsoft Visual C++ Redistributable is not installed, would you like to download it?'), t('Download'), () => {
+                BrowserOpenURL('https://aka.ms/vs/16/release/vc_redist.x64.exe');
+              });
+            } else {
+              toastWithButton(t('Python dependencies are incomplete, would you like to install them?'), t('Install'), () => {
+                InstallPyDep(commonStore.settings.customPythonPath, commonStore.settings.cnMirror);
+                setTimeout(WindowShow, 1000);
+              });
+            }
          } else {
            toast(depErrorMsg, { type: 'error' });
          }
@@ -84,7 +89,8 @@ export const RunButton: FC<{ onClickRun?: MouseEventHandler, iconMode?: boolean
          return;
        }
        commonStore.setDepComplete(true);
-        CopyFile('./backend-python/wkv_cuda_utils/wkv_cuda_model.py', './py310/Lib/site-packages/rwkv/model.py');
+        if (commonStore.platform === 'windows')
+          CopyFile('./backend-python/wkv_cuda_utils/wkv_cuda_model.py', './py310/Lib/site-packages/rwkv/model.py');
        saveCache();
      }

@@ -110,7 +116,7 @@ export const RunButton: FC<{ onClickRun?: MouseEventHandler, iconMode?: boolean

      await exit(1000).catch(() => {
      });
-      StartServer(port, commonStore.settings.host !== '127.0.0.1' ? '0.0.0.0' : '127.0.0.1');
+      StartServer(commonStore.settings.customPythonPath, port, commonStore.settings.host !== '127.0.0.1' ? '0.0.0.0' : '127.0.0.1');
      setTimeout(WindowShow, 1000);

      let timeoutCount = 6;
@@ -136,9 +142,9 @@ export const RunButton: FC<{ onClickRun?: MouseEventHandler, iconMode?: boolean
            });

            let customCudaFile = '';
-            if (modelConfig.modelParameters.useCustomCuda) {
+            if (modelConfig.modelParameters.device != 'CPU' && modelConfig.modelParameters.useCustomCuda) {
              customCudaFile = getSupportedCustomCudaFile();
-              if (customCudaFile) {
+              if (customCudaFile && commonStore.platform === 'windows') {
                FileExists('./py310/Lib/site-packages/rwkv/model.py').then((exist) => {
                  // defensive measure. As Python has already been launched, will only take effect the next time it runs.
                  if (!exist) CopyFile('./backend-python/wkv_cuda_utils/wkv_cuda_model.py', './py310/Lib/site-packages/rwkv/model.py');
@@ -156,7 +162,7 @@ export const RunButton: FC<{ onClickRun?: MouseEventHandler, iconMode?: boolean
            }

            switchModel({
-              model: `${manifest.localModelDir}/${modelConfig.modelParameters.modelName}`,
+              model: modelPath,
              strategy: getStrategy(modelConfig),
              customCuda: customCudaFile !== ''
            }).then((r) => {
@@ -173,7 +179,10 @@ export const RunButton: FC<{ onClickRun?: MouseEventHandler, iconMode?: boolean
              }
            }).catch(() => {
              commonStore.setStatus({ status: ModelStatus.Offline });
-              toast(t('Failed to switch model'), { type: 'error' });
+              if (commonStore.platform === 'windows')
+                toast(t('Failed to switch model, please try starting the program with administrator privileges.'), { type: 'error' });
+              else
+                toast(t('Failed to switch model'), { type: 'error' });
            });
          }
        }).catch(() => {
--- a/frontend/src/pages/Chat.tsx
+++ b/frontend/src/pages/Chat.tsx
@@ -7,7 +7,7 @@ import { v4 as uuid } from 'uuid';
 import classnames from 'classnames';
 import { fetchEventSource } from '@microsoft/fetch-event-source';
 import { ConversationPair, getConversationPairs, Record } from '../utils/get-conversation-pairs';
-import logo from '../../../build/appicon.jpg';
+import logo from '../assets/images/logo.jpg';
 import MarkdownRender from '../components/MarkdownRender';
 import { ToolTipButton } from '../components/ToolTipButton';
 import { ArrowCircleUp28Regular, Delete28Regular, RecordStop28Regular } from '@fluentui/react-icons';
--- a/frontend/src/pages/Completion.tsx
+++ b/frontend/src/pages/Completion.tsx
@@ -101,6 +101,19 @@ export const defaultPresets: CompletionPreset[] = [{
    injectStart: '\\n\\nAlice: ',
    injectEnd: '\\n\\nBob: '
  }
+}, {
+  name: 'Instruction',
+  prompt: 'Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n# Instruction:\nExplain the following metaphor: Life is like cats.\n\n# Response:\n',
+  params: {
+    maxResponseToken: 500,
+    temperature: 1.2,
+    topP: 0.5,
+    presencePenalty: 0.4,
+    frequencyPenalty: 0.4,
+    stop: '',
+    injectStart: '',
+    injectEnd: ''
+  }
 }, {
  name: 'Blank',
  prompt: '',
--- a/frontend/src/pages/Configs.tsx
+++ b/frontend/src/pages/Configs.tsx
@@ -1,5 +1,27 @@
-import { Dropdown, Input, Label, Option, Select, Switch } from '@fluentui/react-components';
-import { AddCircle20Regular, DataUsageSettings20Regular, Delete20Regular, Save20Regular } from '@fluentui/react-icons';
+import {
+  Button,
+  Dialog,
+  DialogActions,
+  DialogBody,
+  DialogContent,
+  DialogSurface,
+  DialogTitle,
+  DialogTrigger,
+  Dropdown,
+  Input,
+  Label,
+  Option,
+  Select,
+  Switch,
+  Text
+} from '@fluentui/react-components';
+import {
+  AddCircle20Regular,
+  ArrowReset20Regular,
+  DataUsageSettings20Regular,
+  Delete20Regular,
+  Save20Regular
+} from '@fluentui/react-icons';
 import React, { FC } from 'react';
 import { Section } from '../components/Section';
 import { Labeled } from '../components/Labeled';
@@ -14,10 +36,11 @@ import { useNavigate } from 'react-router';
 import { RunButton } from '../components/RunButton';
 import { updateConfig } from '../apis';
 import { ConvertModel, FileExists } from '../../wailsjs/go/backend_golang/App';
-import manifest from '../../../manifest.json';
 import { getStrategy, refreshLocalModels } from '../utils';
 import { useTranslation } from 'react-i18next';
 import { WindowShow } from '../../wailsjs/runtime/runtime';
+import strategyImg from '../assets/images/strategy.jpg';
+import strategyZhImg from '../assets/images/strategy_zh.jpg';

 export type ApiParameters = {
  apiPort: number
@@ -28,7 +51,7 @@ export type ApiParameters = {
  frequencyPenalty: number;
 }

-export type Device = 'CPU' | 'CUDA';
+export type Device = 'CPU' | 'CUDA' | 'Custom';
 export type Precision = 'fp16' | 'int8' | 'fp32';

 export type ModelParameters = {
@@ -40,6 +63,7 @@ export type ModelParameters = {
  maxStoredLayers: number;
  enableHighPrecisionForLastLayer: boolean;
  useCustomCuda?: boolean;
+  customStrategy?: string;
 }

 export type ModelConfig = {
@@ -161,7 +185,7 @@ export const defaultModelConfigs: ModelConfig[] = [
      frequencyPenalty: 0.4
    },
    modelParameters: {
-      modelName: 'RWKV-4-Raven-7B-v11-Eng49%-Chn49%-Jpn1%-Other1%-20230430-ctx8192.pth',
+      modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
      device: 'CUDA',
      precision: 'int8',
      storedLayers: 8,
@@ -261,7 +285,7 @@ export const defaultModelConfigs: ModelConfig[] = [
      frequencyPenalty: 0.4
    },
    modelParameters: {
-      modelName: 'RWKV-4-Raven-7B-v11-Eng49%-Chn49%-Jpn1%-Other1%-20230430-ctx8192.pth',
+      modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
      device: 'CUDA',
      precision: 'int8',
      storedLayers: 18,
@@ -341,7 +365,7 @@ export const defaultModelConfigs: ModelConfig[] = [
      frequencyPenalty: 0.4
    },
    modelParameters: {
-      modelName: 'RWKV-4-Raven-7B-v11-Eng49%-Chn49%-Jpn1%-Other1%-20230430-ctx8192.pth',
+      modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
      device: 'CUDA',
      precision: 'int8',
      storedLayers: 27,
@@ -381,7 +405,7 @@ export const defaultModelConfigs: ModelConfig[] = [
      frequencyPenalty: 0.4
    },
    modelParameters: {
-      modelName: 'RWKV-4-Raven-7B-v11-Eng49%-Chn49%-Jpn1%-Other1%-20230430-ctx8192.pth',
+      modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
      device: 'CUDA',
      precision: 'int8',
      storedLayers: 41,
@@ -441,7 +465,7 @@ export const defaultModelConfigs: ModelConfig[] = [
      frequencyPenalty: 0.4
    },
    modelParameters: {
-      modelName: 'RWKV-4-Raven-7B-v11-Eng49%-Chn49%-Jpn1%-Other1%-20230430-ctx8192.pth',
+      modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
      device: 'CUDA',
      precision: 'fp16',
      storedLayers: 41,
@@ -597,7 +621,7 @@ export const defaultModelConfigs: ModelConfig[] = [
      frequencyPenalty: 0.4
    },
    modelParameters: {
-      modelName: 'RWKV-4-Raven-7B-v11-Eng49%-Chn49%-Jpn1%-Other1%-20230430-ctx8192.pth',
+      modelName: 'RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth',
      device: 'CPU',
      precision: 'fp32',
      storedLayers: 41,
@@ -611,6 +635,7 @@ export const Configs: FC = observer(() => {
  const { t } = useTranslation();
  const [selectedIndex, setSelectedIndex] = React.useState(commonStore.currentModelConfigIndex);
  const [selectedConfig, setSelectedConfig] = React.useState(commonStore.modelConfigs[selectedIndex]);
+  const [displayStrategyImg, setDisplayStrategyImg] = React.useState(false);
  const navigate = useNavigate();
  const port = selectedConfig.apiParameters.apiPort;

@@ -679,6 +704,31 @@ export const Configs: FC = observer(() => {
            commonStore.deleteModelConfig(selectedIndex);
            updateSelectedIndex(Math.min(selectedIndex, commonStore.modelConfigs.length - 1));
          }} />
+          <Dialog>
+            <DialogTrigger disableButtonEnhancement>
+              <ToolTipButton desc={t('Reset All Configs')} icon={<ArrowReset20Regular />} />
+            </DialogTrigger>
+            <DialogSurface>
+              <DialogBody>
+                <DialogTitle>{t('Reset All Configs')}</DialogTitle>
+                <DialogContent>
+                  {t('Are you sure you want to reset all configs? This will obtain the latest preset configs, but will override your custom configs and cannot be undone.')}
+                </DialogContent>
+                <DialogActions>
+                  <DialogTrigger disableButtonEnhancement>
+                    <Button appearance="secondary">{t('Cancel')}</Button>
+                  </DialogTrigger>
+                  <DialogTrigger disableButtonEnhancement>
+                    <Button appearance="primary" onClick={() => {
+                      commonStore.setModelConfigs(defaultModelConfigs, false); // updateSelectedIndex() will save configs
+                      updateSelectedIndex(0);
+                    }}>{t('Confirm')}
+                    </Button>
+                  </DialogTrigger>
+                </DialogActions>
+              </DialogBody>
+            </DialogSurface>
+          </Dialog>
          <ToolTipButton desc={t('Save Config')} icon={<Save20Regular />} onClick={onClickSave} />
        </div>
        <div className="flex items-center gap-4">
@@ -789,16 +839,25 @@ export const Configs: FC = observer(() => {
                  </div>
                } />
                <ToolTipButton text={t('Convert')} desc={t('Convert model with these configs')} onClick={async () => {
-                  const modelPath = `${manifest.localModelDir}/${selectedConfig.modelParameters.modelName}`;
+                  if (commonStore.platform == 'darwin') {
+                    toast(t('MacOS is not supported yet, please convert manually.'), { type: 'info' });
+                    return;
+                  }
+
+                  const modelPath = `${commonStore.settings.customModelsPath}/${selectedConfig.modelParameters.modelName}`;
                  if (await FileExists(modelPath)) {
                    const strategy = getStrategy(selectedConfig);
-                    const newModelPath = modelPath + '-' + strategy.replace(/[> *+]/g, '-');
+                    const newModelPath = modelPath + '-' + strategy.replace(/[:> *+]/g, '-');
                    toast(t('Start Converting'), { autoClose: 1000, type: 'info' });
-                    ConvertModel(modelPath, strategy, newModelPath).then(() => {
+                    ConvertModel(commonStore.settings.customPythonPath, modelPath, strategy, newModelPath).then(() => {
                      toast(`${t('Convert Success')} - ${newModelPath}`, { type: 'success' });
                      refreshLocalModels({ models: commonStore.modelSourceList }, false);
                    }).catch(e => {
-                      toast(`${t('Convert Failed')} - ${e.message || e}`, { type: 'error' });
+                      const errMsg = e.message || e;
+                      if (errMsg.includes('path contains space'))
+                        toast(`${t('Convert Failed')} - ${t('Path Cannot Contain Space')}`, { type: 'error' });
+                      else
+                        toast(`${t('Convert Failed')} - ${e.message || e}`, { type: 'error' });
                    });
                    setTimeout(WindowShow, 1000);
                  } else {
@@ -806,69 +865,106 @@ export const Configs: FC = observer(() => {
                  }
                }} />
                <Labeled label={t('Device')} content={
-                  <Dropdown style={{ minWidth: 0 }} className="grow" value={selectedConfig.modelParameters.device}
+                  <Dropdown style={{ minWidth: 0 }} className="grow" value={t(selectedConfig.modelParameters.device)!}
                    selectedOptions={[selectedConfig.modelParameters.device]}
                    onOptionSelect={(_, data) => {
-                      if (data.optionText) {
+                      if (data.optionValue) {
                        setSelectedConfigModelParams({
-                          device: data.optionText as Device
+                          device: data.optionValue as Device
                        });
                      }
                    }}>
-                    <Option>CPU</Option>
-                    <Option>CUDA</Option>
+                    <Option value="CPU">CPU</Option>
+                    <Option value="CUDA">CUDA</Option>
+                    <Option value="Custom">{t('Custom')!}</Option>
                  </Dropdown>
                } />
-                <Labeled label={t('Precision')}
-                  desc={t('int8 uses less VRAM, but has slightly lower quality. fp16 has higher quality, and fp32 has the best quality.')}
-                  content={
-                    <Dropdown style={{ minWidth: 0 }} className="grow"
-                      value={selectedConfig.modelParameters.precision}
-                      selectedOptions={[selectedConfig.modelParameters.precision]}
-                      onOptionSelect={(_, data) => {
-                        if (data.optionText) {
+                {
+                  selectedConfig.modelParameters.device != 'Custom' && <Labeled label={t('Precision')}
+                    desc={t('int8 uses less VRAM, but has slightly lower quality. fp16 has higher quality, and fp32 has the best quality.')}
+                    content={
+                      <Dropdown style={{ minWidth: 0 }} className="grow"
+                        value={selectedConfig.modelParameters.precision}
+                        selectedOptions={[selectedConfig.modelParameters.precision]}
+                        onOptionSelect={(_, data) => {
+                          if (data.optionText) {
+                            setSelectedConfigModelParams({
+                              precision: data.optionText as Precision
+                            });
+                          }
+                        }}>
+                        <Option>fp16</Option>
+                        <Option>int8</Option>
+                        <Option>fp32</Option>
+                      </Dropdown>
+                    } />
+                }
+                {
+                  selectedConfig.modelParameters.device == 'CUDA' &&
+                  <Labeled label={t('Current Strategy')}
+                    content={<Text> {getStrategy(selectedConfig)} </Text>} />
+                }
+                {
+                  selectedConfig.modelParameters.device == 'CUDA' &&
+                  <Labeled label={t('Stored Layers')}
+                    desc={t('Number of the neural network layers loaded into VRAM, the more you load, the faster the speed, but it consumes more VRAM.')}
+                    content={
+                      <ValuedSlider value={selectedConfig.modelParameters.storedLayers} min={0}
+                        max={selectedConfig.modelParameters.maxStoredLayers} step={1} input
+                        onChange={(e, data) => {
                          setSelectedConfigModelParams({
-                            precision: data.optionText as Precision
+                            storedLayers: data.value
                          });
-                        }
-                      }}>
-                      <Option>fp16</Option>
-                      <Option>int8</Option>
-                      <Option>fp32</Option>
-                    </Dropdown>
-                  } />
-                <div />
-                <Labeled label={t('Stored Layers')}
-                  desc={t('Number of the neural network layers loaded into VRAM, the more you load, the faster the speed, but it consumes more VRAM.')}
-                  content={
-                    <ValuedSlider value={selectedConfig.modelParameters.storedLayers} min={0}
-                      max={selectedConfig.modelParameters.maxStoredLayers} step={1} input
-                      onChange={(e, data) => {
-                        setSelectedConfigModelParams({
-                          storedLayers: data.value
-                        });
-                      }} />
-                  } />
-                <Labeled label={t('Enable High Precision For Last Layer')}
-                  desc={t('Whether to use CPU to calculate the last output layer of the neural network with FP32 precision to obtain better quality.')}
-                  content={
-                    <Switch checked={selectedConfig.modelParameters.enableHighPrecisionForLastLayer}
-                      onChange={(e, data) => {
-                        setSelectedConfigModelParams({
-                          enableHighPrecisionForLastLayer: data.checked
-                        });
-                      }} />
-                  } />
-                <Labeled label={t('Use Custom CUDA kernel to Accelerate')}
-                  desc={t('Enabling this option can greatly improve inference speed, but there may be compatibility issues. If it fails to start, please turn off this option.')}
-                  content={
-                    <Switch checked={selectedConfig.modelParameters.useCustomCuda}
-                      onChange={(e, data) => {
-                        setSelectedConfigModelParams({
-                          useCustomCuda: data.checked
-                        });
-                      }} />
-                  } />
+                        }} />
+                    } />
+                }
+                {
+                  selectedConfig.modelParameters.device == 'CUDA' &&
+                  <Labeled label={t('Enable High Precision For Last Layer')}
+                    desc={t('Whether to use CPU to calculate the last output layer of the neural network with FP32 precision to obtain better quality.')}
+                    content={
+                      <Switch checked={selectedConfig.modelParameters.enableHighPrecisionForLastLayer}
+                        onChange={(e, data) => {
+                          setSelectedConfigModelParams({
+                            enableHighPrecisionForLastLayer: data.checked
+                          });
+                        }} />
+                    } />
+                }
+                {
+                  displayStrategyImg &&
+                  <img style={{ width: '80vh', height: 'auto', zIndex: 100 }} className="fixed left-0 top-0"
+                    src={commonStore.settings.language === 'zh' ? strategyZhImg : strategyImg} />
+                }
+                {
+                  selectedConfig.modelParameters.device == 'Custom' &&
+                  <Labeled label="Strategy"
+                    onMouseEnter={() => setDisplayStrategyImg(true)}
+                    onMouseLeave={() => setDisplayStrategyImg(false)}
+                    content={
+                      <Input className="grow" placeholder="cuda:0 fp16 *20 -> cuda:1 fp16"
+                        value={selectedConfig.modelParameters.customStrategy}
+                        onChange={(e, data) => {
+                          setSelectedConfigModelParams({
+                            customStrategy: data.value
+                          });
+                        }} />
+                    } />
+                }
+                {selectedConfig.modelParameters.device == 'Custom' && <div />}
+                {
+                  selectedConfig.modelParameters.device != 'CPU' &&
+                  <Labeled label={t('Use Custom CUDA kernel to Accelerate')}
+                    desc={t('Enabling this option can greatly improve inference speed, but there may be compatibility issues. If it fails to start, please turn off this option.')}
+                    content={
+                      <Switch checked={selectedConfig.modelParameters.useCustomCuda}
+                        onChange={(e, data) => {
+                          setSelectedConfigModelParams({
+                            useCustomCuda: data.checked
+                          });
+                        }} />
+                    } />
+                }
              </div>
            }
          />
--- a/frontend/src/pages/Downloads.tsx
+++ b/frontend/src/pages/Downloads.tsx
@@ -62,7 +62,7 @@ export const Downloads: FC = observer(() => {
                        ContinueDownload(status.url);
                    }} />}
                <ToolTipButton desc={t('Open Folder')} icon={<Folder20Regular />} onClick={() => {
-                  OpenFileFolder(status.path);
+                  OpenFileFolder(`${commonStore.settings.customModelsPath}/${status.name}`);
                }} />
              </div>
            </Field>
--- a/frontend/src/pages/Models.tsx
+++ b/frontend/src/pages/Models.tsx
@@ -18,7 +18,6 @@ import { observer } from 'mobx-react-lite';
 import commonStore from '../stores/commonStore';
 import { BrowserOpenURL } from '../../wailsjs/runtime';
 import { AddToDownloadList, OpenFileFolder } from '../../wailsjs/go/backend_golang/App';
-import manifest from '../../../manifest.json';
 import { Page } from '../components/Page';
 import { bytesToGb, refreshModels, saveConfigs, toastWithButton } from '../utils';
 import { useTranslation } from 'react-i18next';
@@ -143,7 +142,7 @@ const columns: TableColumnDefinition<ModelSourceItem>[] = [
            {
              item.isLocal &&
              <ToolTipButton desc={t('Open Folder')} icon={<Folder20Regular />} onClick={() => {
-                OpenFileFolder(`./${manifest.localModelDir}/${item.name}`);
+                OpenFileFolder(`${commonStore.settings.customModelsPath}/${item.name}`);
              }} />
            }
            {item.downloadUrl && !item.isLocal &&
@@ -152,7 +151,7 @@ const columns: TableColumnDefinition<ModelSourceItem>[] = [
                    navigate({ pathname: '/downloads' });
                  },
                  { autoClose: 3000 });
-                AddToDownloadList(`./${manifest.localModelDir}/${item.name}`, item.downloadUrl!);
+                AddToDownloadList(`${commonStore.settings.customModelsPath}/${item.name}`, item.downloadUrl!);
              }} />}
            {item.url && <ToolTipButton desc={t('Open Url')} icon={<Open20Regular />} onClick={() => {
              BrowserOpenURL(item.url!);
--- a/frontend/src/pages/Settings.tsx
+++ b/frontend/src/pages/Settings.tsx
@@ -1,6 +1,15 @@
-import React, { FC } from 'react';
+import React, { FC, useEffect, useRef } from 'react';
 import { Page } from '../components/Page';
-import { Dropdown, Option, Switch } from '@fluentui/react-components';
+import {
+  Accordion,
+  AccordionHeader,
+  AccordionItem,
+  AccordionPanel,
+  Dropdown,
+  Input,
+  Option,
+  Switch
+} from '@fluentui/react-components';
 import { Labeled } from '../components/Labeled';
 import commonStore from '../stores/commonStore';
 import { observer } from 'mobx-react-lite';
@@ -15,16 +24,24 @@ export const Languages = {
 export type Language = keyof typeof Languages;

 export type SettingsType = {
-  language: Language,
+  language: Language
  darkMode: boolean
  autoUpdatesCheck: boolean
  giteeUpdatesSource: boolean
  cnMirror: boolean
  host: string
+  customModelsPath: string
+  customPythonPath: string
 }

 export const Settings: FC = observer(() => {
  const { t, i18n } = useTranslation();
+  const advancedHeaderRef = useRef<HTMLDivElement>(null);
+
+  useEffect(() => {
+    if (advancedHeaderRef.current)
+      (advancedHeaderRef.current.firstElementChild as HTMLElement).style.padding = '0';
+  }, []);

  return (
    <Page title={t('Settings')} content={
@@ -96,6 +113,35 @@ export const Settings: FC = observer(() => {
              });
            }} />
        } />
+        <Accordion collapsible>
+          <AccordionItem value="1">
+            <AccordionHeader ref={advancedHeaderRef} size="large">{t('Advanced')}</AccordionHeader>
+            <AccordionPanel>
+              <div className="flex flex-col gap-2 overflow-hidden">
+                {commonStore.platform !== 'darwin' &&
+                  <Labeled label={t('Custom Models Path')}
+                    content={
+                      <Input className="grow" placeholder="./models" value={commonStore.settings.customModelsPath}
+                        onChange={(e, data) => {
+                          commonStore.setSettings({
+                            customModelsPath: data.value
+                          });
+                        }} />
+                    } />
+                }
+                <Labeled label={t('Custom Python Path')} // if set, will not use precompiled cuda kernel
+                  content={
+                    <Input className="grow" placeholder="./py310/python" value={commonStore.settings.customPythonPath}
+                      onChange={(e, data) => {
+                        commonStore.setSettings({
+                          customPythonPath: data.value
+                        });
+                      }} />
+                  } />
+              </div>
+            </AccordionPanel>
+          </AccordionItem>
+        </Accordion>
      </div>
    } />
  );
--- a/frontend/src/startup.ts
+++ b/frontend/src/startup.ts
@@ -1,4 +1,4 @@
-import commonStore from './stores/commonStore';
+import commonStore, { Platform } from './stores/commonStore';
 import { GetPlatform, ReadJson } from '../wailsjs/go/backend_golang/App';
 import { Cache, checkUpdate, downloadProgramFiles, LocalConfig, refreshModels } from './utils';
 import { getStatus } from './apis';
@@ -13,11 +13,11 @@ export async function startup() {
      commonStore.setDownloadList(data);
  });

-  initCache().then(initRemoteText);
-
-  await GetPlatform().then(p => commonStore.setPlatform(p));
+  await GetPlatform().then(p => commonStore.setPlatform(p as Platform));
  await initConfig();

+  initCache().then(initRemoteText); // depends on config customModelsPath
+
  if (commonStore.settings.autoUpdatesCheck) // depends on config settings
    checkUpdate();

--- a/frontend/src/stores/commonStore.ts
+++ b/frontend/src/stores/commonStore.ts
@@ -25,6 +25,8 @@ export type Status = {
  device_name: string;
 }

+export type Platform = 'windows' | 'darwin' | 'linux';
+
 class CommonStore {
  // global
  status: Status = {
@@ -33,7 +35,7 @@ class CommonStore {
    device_name: 'CPU'
  };
  depComplete: boolean = false;
-  platform: string = 'windows';
+  platform: Platform = 'windows';
  // home
  introduction: IntroductionContent = manifest.introduction;
  // chat
@@ -57,7 +59,9 @@ class CommonStore {
    autoUpdatesCheck: true,
    giteeUpdatesSource: getUserLanguage() === 'zh',
    cnMirror: getUserLanguage() === 'zh',
-    host: '127.0.0.1'
+    host: '127.0.0.1',
+    customModelsPath: './models',
+    customPythonPath: ''
  };
  // about
  about: AboutContent = manifest.about;
@@ -87,14 +91,16 @@ class CommonStore {
  };

  setModelConfigs = (configs: ModelConfig[], saveConfig: boolean = true) => {
-    this.modelConfigs = configs;
+    this.modelConfigs = JSON.parse(JSON.stringify(configs)); // deep copy
    if (saveConfig)
      saveConfigs();
  };

  createModelConfig = (config: ModelConfig = defaultModelConfigs[0], saveConfig: boolean = true) => {
-    if (config.name === defaultModelConfigs[0].name)
+    if (config.name === defaultModelConfigs[0].name) {
+      config = JSON.parse(JSON.stringify(config)); // deep copy
      config.name = new Date().toLocaleString();
+    }
    this.modelConfigs.push(config);
    if (saveConfig)
      saveConfigs();
@@ -170,7 +176,7 @@ class CommonStore {
    this.completionGenerating = value;
  }

-  setPlatform(value: string) {
+  setPlatform(value: Platform) {
    this.platform = value;
  }
 }
--- a/frontend/src/utils/index.tsx
+++ b/frontend/src/utils/index.tsx
@@ -51,7 +51,7 @@ export async function refreshLocalModels(cache: { models: ModelSourceItem[] }, f
  if (filter)
    cache.models = cache.models.filter(m => !m.isLocal); //TODO BUG cause local but in manifest files to be removed, so currently cache is disabled

-  await ListDirFiles(manifest.localModelDir).then((data) => {
+  await ListDirFiles(commonStore.settings.customModelsPath).then((data) => {
    cache.models.push(...data.flatMap(d => {
      if (!d.isDir && d.name.endsWith('.pth'))
        return [{
@@ -127,12 +127,23 @@ export const getStrategy = (modelConfig: ModelConfig | undefined = undefined) =>
  if (modelConfig) params = modelConfig.modelParameters;
  else params = commonStore.getCurrentModelConfig().modelParameters;
  let strategy = '';
-  strategy += (params.device === 'CPU' ? 'cpu' : 'cuda') + ' ';
-  strategy += params.device === 'CPU' ? 'fp32' : (params.precision === 'fp16' ? 'fp16' : params.precision === 'int8' ? 'fp16i8' : 'fp32');
-  if (params.storedLayers < params.maxStoredLayers)
-    strategy += ` *${params.storedLayers}+`;
-  if (params.enableHighPrecisionForLastLayer)
-    strategy += ' -> cpu fp32 *1';
+  switch (params.device) {
+    case 'CPU':
+      strategy += 'cpu ';
+      strategy += params.precision === 'int8' ? 'fp32i8' : 'fp32';
+      break;
+    case 'CUDA':
+      strategy += 'cuda ';
+      strategy += params.precision === 'fp16' ? 'fp16' : params.precision === 'int8' ? 'fp16i8' : 'fp32';
+      if (params.storedLayers < params.maxStoredLayers)
+        strategy += ` *${params.storedLayers}+`;
+      if (params.enableHighPrecisionForLastLayer)
+        strategy += ' -> cpu fp32 *1';
+      break;
+    case 'Custom':
+      strategy = params.customStrategy || '';
+      break;
+  }
  return strategy;
 };

@@ -298,9 +309,9 @@ export function toastWithButton(text: string, buttonText: string, onClickButton:
 }

 export function getSupportedCustomCudaFile() {
-  if ([' 10', ' 16', ' 20', ' 30', 'P40'].some(v => commonStore.status.device_name.includes(v)))
+  if ([' 10', ' 16', ' 20', ' 30', 'MX', 'Tesla P', 'Quadro P', 'NVIDIA P', 'TITAN X', 'TITAN RTX', 'RTX A'].some(v => commonStore.status.device_name.includes(v)))
    return './backend-python/wkv_cuda_utils/wkv_cuda10_30.pyd';
-  else if ([' 40'].some(v => commonStore.status.device_name.includes(v)))
+  else if ([' 40', 'RTX TITAN Ada'].some(v => commonStore.status.device_name.includes(v)))
    return './backend-python/wkv_cuda_utils/wkv_cuda40.pyd';
  else
    return '';
--- a/frontend/wailsjs/go/backend_golang/App.d.ts
+++ b/frontend/wailsjs/go/backend_golang/App.d.ts
@@ -6,13 +6,13 @@ export function AddToDownloadList(arg1:string,arg2:string):Promise<void>;

 export function ContinueDownload(arg1:string):Promise<void>;

-export function ConvertModel(arg1:string,arg2:string,arg3:string):Promise<string>;
+export function ConvertModel(arg1:string,arg2:string,arg3:string,arg4:string):Promise<string>;

 export function CopyFile(arg1:string,arg2:string):Promise<void>;

 export function DeleteFile(arg1:string):Promise<void>;

-export function DepCheck():Promise<void>;
+export function DepCheck(arg1:string):Promise<void>;

 export function DownloadFile(arg1:string,arg2:string):Promise<void>;

@@ -20,7 +20,7 @@ export function FileExists(arg1:string):Promise<boolean>;

 export function GetPlatform():Promise<string>;

-export function InstallPyDep(arg1:boolean):Promise<string>;
+export function InstallPyDep(arg1:string,arg2:boolean):Promise<string>;

 export function ListDirFiles(arg1:string):Promise<Array<backend_golang.FileInfo>>;

@@ -34,6 +34,6 @@ export function ReadJson(arg1:string):Promise<any>;

 export function SaveJson(arg1:string,arg2:any):Promise<void>;

-export function StartServer(arg1:number,arg2:string):Promise<string>;
+export function StartServer(arg1:string,arg2:number,arg3:string):Promise<string>;

 export function UpdateApp(arg1:string):Promise<boolean>;
--- a/frontend/wailsjs/go/backend_golang/App.js
+++ b/frontend/wailsjs/go/backend_golang/App.js
@@ -10,8 +10,8 @@ export function ContinueDownload(arg1) {
  return window['go']['backend_golang']['App']['ContinueDownload'](arg1);
 }

-export function ConvertModel(arg1, arg2, arg3) {
-  return window['go']['backend_golang']['App']['ConvertModel'](arg1, arg2, arg3);
+export function ConvertModel(arg1, arg2, arg3, arg4) {
+  return window['go']['backend_golang']['App']['ConvertModel'](arg1, arg2, arg3, arg4);
 }

 export function CopyFile(arg1, arg2) {
@@ -22,8 +22,8 @@ export function DeleteFile(arg1) {
  return window['go']['backend_golang']['App']['DeleteFile'](arg1);
 }

-export function DepCheck() {
-  return window['go']['backend_golang']['App']['DepCheck']();
+export function DepCheck(arg1) {
+  return window['go']['backend_golang']['App']['DepCheck'](arg1);
 }

 export function DownloadFile(arg1, arg2) {
@@ -38,8 +38,8 @@ export function GetPlatform() {
  return window['go']['backend_golang']['App']['GetPlatform']();
 }

-export function InstallPyDep(arg1) {
-  return window['go']['backend_golang']['App']['InstallPyDep'](arg1);
+export function InstallPyDep(arg1, arg2) {
+  return window['go']['backend_golang']['App']['InstallPyDep'](arg1, arg2);
 }

 export function ListDirFiles(arg1) {
@@ -66,8 +66,8 @@ export function SaveJson(arg1, arg2) {
  return window['go']['backend_golang']['App']['SaveJson'](arg1, arg2);
 }

-export function StartServer(arg1, arg2) {
-  return window['go']['backend_golang']['App']['StartServer'](arg1, arg2);
+export function StartServer(arg1, arg2, arg3) {
+  return window['go']['backend_golang']['App']['StartServer'](arg1, arg2, arg3);
 }

 export function UpdateApp(arg1) {
--- a/frontend/wailsjs/go/models.ts
+++ b/frontend/wailsjs/go/models.ts
--- a/main.go
+++ b/main.go
@@ -2,6 +2,8 @@ package main

 import (
 	"embed"
+	"runtime/debug"
+	"strings"

 	backend "rwkv-runner/backend-golang"

@@ -23,9 +25,11 @@ var cyacInfo embed.FS
 var py embed.FS

 func main() {
-	backend.CopyEmbed(cyac)
-	backend.CopyEmbed(cyacInfo)
-	backend.CopyEmbed(py)
+	if buildInfo, ok := debug.ReadBuildInfo(); !ok || strings.Contains(buildInfo.String(), "-ldflags") {
+		backend.CopyEmbed(cyac)
+		backend.CopyEmbed(cyacInfo)
+		backend.CopyEmbed(py)
+	}

 	// Create an instance of the app structure
 	app := backend.NewApp()
@@ -34,7 +38,7 @@ func main() {
 	err := wails.Run(&options.App{
 		Title:     "RWKV-Runner",
 		Width:     1024,
-		Height:    640,
+		Height:    680,
 		MinWidth:  375,
 		MinHeight: 640,
 		AssetServer: &assetserver.Options{
--- a/manifest.json
+++ b/manifest.json
@@ -1,5 +1,5 @@
 {
-  "version": "1.1.2",
+  "version": "1.1.6",
  "introduction": {
    "en": "RWKV is an open-source, commercially usable large language model with high flexibility and great potential for development.\n### About This Tool\nThis tool aims to lower the barrier of entry for using large language models, making it accessible to everyone. It provides fully automated dependency and model management. You simply need to click and run, following the instructions, to deploy a local large language model. The tool itself is very compact and only requires a single executable file for one-click deployment.\nAdditionally, this tool offers an interface that is fully compatible with the OpenAI API. This means you can use any ChatGPT client as a client for RWKV, enabling capability expansion beyond just chat functionality.\n### Preset Configuration Rules at the Bottom\nThis tool comes with a series of preset configurations to reduce complexity. The naming rules for each configuration represent the following in order: device - required VRAM/memory - model size - model language.\nFor example, \"GPU-8G-3B-EN\" indicates that this configuration is for a graphics card with 8GB of VRAM, a model size of 3 billion parameters, and it uses an English language model.\nLarger model sizes have higher performance and VRAM requirements. Among configurations with the same model size, those with higher VRAM usage will have faster runtime.\nFor example, if you have 12GB of VRAM but running the \"GPU-12G-7B-EN\" configuration is slow, you can downgrade to \"GPU-8G-3B-EN\" for a significant speed improvement.\n### About RWKV\nRWKV is an RNN with Transformer-level LLM performance, which can also be directly trained like a GPT transformer (parallelizable). And it's 100% attention-free. You only need the hidden state at position t to compute the state at position t+1. You can use the \"GPT\" mode to quickly compute the hidden state for the \"RNN\" mode.<br/>So it's combining the best of RNN and transformer - great performance, fast inference, saves VRAM, fast training, \"infinite\" ctx_len, and free sentence embedding (using the final hidden state).",
    "zh": "RWKV是一个开源且允许商用的大语言模型，灵活性很高且极具发展潜力。\n### 关于本工具\n本工具旨在降低大语言模型的使用门槛，做到人人可用，本工具提供了全自动化的依赖和模型管理，你只需要直接点击运行，跟随引导，即可完成本地大语言模型的部署，工具本身体积极小，只需要一个exe即可完成一键部署。\n此外，本工具提供了与OpenAI API完全兼容的接口，这意味着你可以把任意ChatGPT客户端用作RWKV的客户端，实现能力拓展，而不局限于聊天。\n### 底部的预设配置规则\n本工具内置了一系列预设配置，以降低使用难度，每个配置名的规则，依次代表着：设备-所需显存/内存-模型规模-模型语言。\n例如，GPU-8G-3B-CN，表示该配置用于显卡，需要8G显存，模型规模为30亿参数，使用的是中文模型。\n模型规模越大，性能要求越高，显存要求也越高，而同样模型规模的配置中，显存占用越高的，运行速度越快。\n例如当你有12G显存，但运行GPU-12G-7B-CN配置速度比较慢，可降级成GPU-8G-3B-CN，将会大幅提速。\n### 关于RWKV\nRWKV是具有Transformer级别LLM性能的RNN，也可以像GPT Transformer一样直接进行训练（可并行化）。而且它是100% attention-free的。你只需在位置t处获得隐藏状态即可计算位置t + 1处的状态。你可以使用“GPT”模式快速计算用于“RNN”模式的隐藏状态。\n因此，它将RNN和Transformer的优点结合起来 - 高性能、快速推理、节省显存、快速训练、“无限”上下文长度以及免费的语句嵌入（使用最终隐藏状态）。"
@@ -8,7 +8,6 @@
    "en": "<div align=\"center\">\n\nProject Source Code:\nhttps://github.com/josStorer/RWKV-Runner\nAuthor: [@josStorer](https://github.com/josStorer)\nFAQs: https://github.com/josStorer/RWKV-Runner/wiki/FAQs\n\nRelated Repositories:\nRWKV-4-Raven: https://huggingface.co/BlinkDL/rwkv-4-raven/tree/main\nChatRWKV: https://github.com/BlinkDL/ChatRWKV\nRWKV-LM: https://github.com/BlinkDL/RWKV-LM\n\n</div>",
    "zh": "<div align=\"center\">\n\n本项目源码:\nhttps://github.com/josStorer/RWKV-Runner\n作者: [@josStorer](https://github.com/josStorer)\n演示与常见问题说明视频: https://www.bilibili.com/video/BV1hM4y1v76R\n疑难解答: https://www.bilibili.com/read/cv23921171\n\n相关仓库:\nRWKV-4-Raven: https://huggingface.co/BlinkDL/rwkv-4-raven/tree/main\nChatRWKV: https://github.com/BlinkDL/ChatRWKV\nRWKV-LM: https://github.com/BlinkDL/RWKV-LM\n\n</div>"
  },
-  "localModelDir": "models",
  "programFiles": [
    {
      "url": "",
@@ -76,7 +75,20 @@
      "SHA256": "9e67a74964abcb4463711e447ddf47735561d7b40592d2d02b29d2e796a4fd14",
      "lastUpdated": "2023-04-30T14:35:59",
      "url": "https://huggingface.co/BlinkDL/rwkv-4-raven/blob/main/RWKV-4-Raven-7B-v11-Eng49%25-Chn49%25-Jpn1%25-Other1%25-20230430-ctx8192.pth",
-      "downloadUrl": "https://huggingface.co/BlinkDL/rwkv-4-raven/resolve/main/RWKV-4-Raven-7B-v11-Eng49%25-Chn49%25-Jpn1%25-Other1%25-20230430-ctx8192.pth"
+      "downloadUrl": "https://huggingface.co/BlinkDL/rwkv-4-raven/resolve/main/RWKV-4-Raven-7B-v11-Eng49%25-Chn49%25-Jpn1%25-Other1%25-20230430-ctx8192.pth",
+      "hide": true
+    },
+    {
+      "name": "RWKV-4-Raven-7B-v12-Eng49%-Chn49%-Jpn1%-Other1%-20230530-ctx8192.pth",
+      "desc": {
+        "en": "Chinese 7B v12",
+        "zh": "中文 7B v12"
+      },
+      "size": 14785389874,
+      "SHA256": "6d4a089ff36d5d9d96b669d425fc5e4e3959cab426535b52e2364df08f58b407",
+      "lastUpdated": "2023-05-30T23:16:12",
+      "url": "https://huggingface.co/BlinkDL/rwkv-4-raven/blob/main/RWKV-4-Raven-7B-v12-Eng49%25-Chn49%25-Jpn1%25-Other1%25-20230530-ctx8192.pth",
+      "downloadUrl": "https://huggingface.co/BlinkDL/rwkv-4-raven/resolve/main/RWKV-4-Raven-7B-v12-Eng49%25-Chn49%25-Jpn1%25-Other1%25-20230530-ctx8192.pth"
    },
    {
      "name": "RWKV-4-Raven-7B-v12-Eng98%-Other2%-20230521-ctx8192.pth",
@@ -226,6 +238,18 @@
      "url": "https://huggingface.co/BlinkDL/rwkv-4-novel/blob/main/RWKV-4-Novel-7B-v1-ChnEng-20230426-ctx8192.pth",
      "downloadUrl": "https://huggingface.co/BlinkDL/rwkv-4-novel/resolve/main/RWKV-4-Novel-7B-v1-ChnEng-20230426-ctx8192.pth"
    },
+    {
+      "name": "RWKV-4-World-7B-v1-OnlyForTest_40%_trained-20230601-ctx4096.pth",
+      "desc": {
+        "en": "100+ Languages 7B v1 Test",
+        "zh": "100+ 语言 7B v1 测试"
+      },
+      "size": 15035393581,
+      "SHA256": "63c060c472e45b6c3af2baaaee448ffd95f9b46e3cc6e1ef70ce7ecb1d01bcfa",
+      "lastUpdated": "2023-06-02T00:09:39",
+      "url": "https://huggingface.co/BlinkDL/rwkv-4-world/blob/main/RWKV-4-World-7B-v1-OnlyForTest_40%25_trained-20230601-ctx4096.pth",
+      "downloadUrl": "https://huggingface.co/BlinkDL/rwkv-4-world/resolve/main/RWKV-4-World-7B-v1-OnlyForTest_40%25_trained-20230601-ctx4096.pth"
+    },
    {
      "name": "RWKV-4-World-7B-v1-OnlyForTest_30%_trained-20230529-ctx4096.pth",
      "desc": {
@@ -236,7 +260,8 @@
      "SHA256": "05f91562b2ae8b025226e40b3fb536d6f8eb3c142ac899c0808ee1c9dc189ec4",
      "lastUpdated": "2023-05-29T13:25:53",
      "url": "https://huggingface.co/BlinkDL/rwkv-4-world/blob/main/RWKV-4-World-7B-v1-OnlyForTest_30%25_trained-20230529-ctx4096.pth",
-      "downloadUrl": "https://huggingface.co/BlinkDL/rwkv-4-world/resolve/main/RWKV-4-World-7B-v1-OnlyForTest_30%25_trained-20230529-ctx4096.pth"
+      "downloadUrl": "https://huggingface.co/BlinkDL/rwkv-4-world/resolve/main/RWKV-4-World-7B-v1-OnlyForTest_30%25_trained-20230529-ctx4096.pth",
+      "hide": true
    },
    {
      "name": "RWKV-4-World-0.1B-v1-20230520-ctx4096.pth",
--- a/vendor.yml
+++ b/vendor.yml
@@ -0,0 +1,6 @@
+- ^backend-python/rwkv_pip/
+- ^backend-python/wkv_cuda_utils/
+- ^backend-python/get-pip\.py
+- ^backend-python/convert_model\.py
+- ^build/
+- ^frontend/wailsjs/
Author	SHA1	Message	Date
josc146	edc6ac7297	chore	2023-06-03 20:34:33 +08:00
josc146	e89e23621c	update readme	2023-06-03 20:28:21 +08:00
josc146	6b9ec4c6fa	add strategy guides	2023-06-03 20:18:57 +08:00
josc146	ced0966ffc	display current strategy	2023-06-03 19:38:24 +08:00
josc146	966b912013	improve logs	2023-06-03 19:28:37 +08:00
josc146	dc71054e61	improve logs	2023-06-03 17:36:50 +08:00
josc146	408f3c1a4d	release v1.1.6	2023-06-03 17:15:11 +08:00
josc146	38b775c937	add logs	2023-06-03 17:12:59 +08:00
josc146	f2ec1067bf	MX, Tesla P, Quadro P, NVIDIA P, TITAN X, RTX A series, TITAN RTX and RTX TITAN Ada	2023-06-03 12:46:56 +08:00
josc146	b01584c49e	chore	2023-06-03 00:10:31 +08:00
josc146	01e56382a3	release v1.1.5	2023-06-02 23:53:37 +08:00
josc146	391c067250	add Instruction to Completion Presets	2023-06-02 23:53:25 +08:00
josc146	5b98b5c0a7	update manifest.json	2023-06-02 23:45:37 +08:00
josc146	5bde0abb8d	reminder to use administrator privileges	2023-06-02 23:42:13 +08:00
josc146	1036852924	add path contains space prompt and chore	2023-06-02 23:35:33 +08:00
josc146	2b10ccd507	add vc++ installation guide	2023-06-02 23:32:47 +08:00
josc146	e1df1bfc3f	chore	2023-06-02 22:20:57 +08:00
josc146	b41a2e7039	move state cache to memory (todo: state cache db)	2023-06-02 21:33:57 +08:00
josc146	b63370928d	macOS	2023-06-01 16:54:21 +08:00
josc146	06a125b8d7	release v1.1.4	2023-05-31 16:27:52 +08:00
josc146	b03b00419b	fix Cmd and CopyFile	2023-05-31 16:27:43 +08:00
josc146	2f5a7d2d51	fix_tokens	2023-05-31 16:07:09 +08:00
josc146	e318331909	release v1.1.3	2023-05-31 15:46:49 +08:00
josc146	796e5f22c0	custom python path	2023-05-31 15:45:26 +08:00
josc146	b49968c145	custom models path	2023-05-31 15:21:47 +08:00
josc146	cf16e54463	fix_tokens	2023-05-31 14:55:13 +08:00
josc146	7bc8da2e29	add button to reset all configs	2023-05-31 14:19:19 +08:00
josc146	26174d4c10	chore	2023-05-31 14:14:25 +08:00
josc146	3c18ce34c7	cn 7b v12	2023-05-31 12:53:05 +08:00
josc146	c8b2bb53ef	improve system for rwkv-4-world	2023-05-31 12:46:06 +08:00
josc146	9f5d15a7d5	custom strategy mode	2023-05-31 12:26:10 +08:00
josc146	8291c50058	safe ModelConfigBody	2023-05-30 23:13:27 +08:00
josc146	1f3f6cf9a8	chore	2023-05-30 22:58:52 +08:00
josc146	83905c65c7	tesla P104 P106 kernel	2023-05-30 22:46:53 +08:00
josc146	9af2b1d208	update readme	2023-05-30 14:40:33 +08:00
josc146	67c9381b82	upload .gitattributes	2023-05-30 13:17:45 +08:00
josc146	4bdfa2d54c	update readme	2023-05-30 11:52:38 +08:00
josc146	9945338458	chore	2023-05-30 11:52:33 +08:00
josc146	d93157bde4	don't release embedded files in development mode	2023-05-30 11:04:11 +08:00
josc146	4db9b13803	upload vendor.yml	2023-05-30 10:35:24 +08:00