release v1.3.5

.github/workflows/release.yml

@@ -83,6 +83,9 @@ jobs:
           go install github.com/wailsapp/wails/v2/cmd/wails@latest
           rm -rf ./backend-python/wkv_cuda_utils
           rm ./backend-python/get-pip.py
+          sed -i '1,2d' ./backend-golang/wsl_not_windows.go
+          rm ./backend-golang/wsl.go
+          mv ./backend-golang/wsl_not_windows.go ./backend-golang/wsl.go
           make
           mv build/bin/RWKV-Runner build/bin/RWKV-Runner_linux_x64
 
@@ -102,6 +105,9 @@ jobs:
           go install github.com/wailsapp/wails/v2/cmd/wails@latest
           rm -rf ./backend-python/wkv_cuda_utils
           rm ./backend-python/get-pip.py
+          sed -i '' '1,2d' ./backend-golang/wsl_not_windows.go
+          rm ./backend-golang/wsl.go
+          mv ./backend-golang/wsl_not_windows.go ./backend-golang/wsl.go
           make
           cp build/darwin/Readme_Install.txt build/bin/Readme_Install.txt
           cp build/bin/RWKV-Runner.app/Contents/MacOS/RWKV-Runner build/bin/RWKV-Runner_darwin_universal

CURRENT_CHANGE.md

@@ -1,6 +1,13 @@
 ## Changes
 
-- improve lora finetune process (need to be refactored)
+- fix build for macos and linux
+- update Related Repositories
+- fix loss parser
+- improve wsl dependencies installation
+- improve finetune guide
+- refresh local models in real-time (#98)
+- improve python script error messages
+- support using directory as training data
 
 ## Install
 

README.md

@@ -136,9 +136,11 @@ for i in np.argsort(embeddings_cos_sim)[::-1]:
 
 ## Related Repositories:
 
+- RWKV-4-World: https://huggingface.co/BlinkDL/rwkv-4-world/tree/main
 - RWKV-4-Raven: https://huggingface.co/BlinkDL/rwkv-4-raven/tree/main
 - ChatRWKV: https://github.com/BlinkDL/ChatRWKV
 - RWKV-LM: https://github.com/BlinkDL/RWKV-LM
+- RWKV-LM-LoRA: https://github.com/Blealtan/RWKV-LM-LoRA
 
 ## Preview
 

README_JA.md

@@ -126,9 +126,11 @@ for i in np.argsort(embeddings_cos_sim)[::-1]:
 
 ## 関連リポジトリ:
 
+- RWKV-4-World: https://huggingface.co/BlinkDL/rwkv-4-world/tree/main
 - RWKV-4-Raven: https://huggingface.co/BlinkDL/rwkv-4-raven/tree/main
 - ChatRWKV: https://github.com/BlinkDL/ChatRWKV
 - RWKV-LM: https://github.com/BlinkDL/RWKV-LM
+- RWKV-LM-LoRA: https://github.com/Blealtan/RWKV-LM-LoRA
 
 ## プレビュー
 

README_ZH.md

@@ -136,9 +136,11 @@ for i in np.argsort(embeddings_cos_sim)[::-1]:
 
 ## 相关仓库:
 
+- RWKV-4-World: https://huggingface.co/BlinkDL/rwkv-4-world/tree/main
 - RWKV-4-Raven: https://huggingface.co/BlinkDL/rwkv-4-raven/tree/main
 - ChatRWKV: https://github.com/BlinkDL/ChatRWKV
 - RWKV-LM: https://github.com/BlinkDL/RWKV-LM
+- RWKV-LM-LoRA: https://github.com/Blealtan/RWKV-LM-LoRA
 
 ## Preview
 

backend-golang/rwkv.go

@@ -43,6 +43,52 @@ func (a *App) ConvertData(python string, input string, outputPrefix string, voca
 	if strings.Contains(vocab, "rwkv_vocab_v20230424") {
 		tokenizerType = "RWKVTokenizer"
 	}
+
+	input = strings.TrimSuffix(input, "/")
+	if fi, err := os.Stat(input); err == nil && fi.IsDir() {
+		files, err := os.ReadDir(input)
+		if err != nil {
+			return "", err
+		}
+		jsonlFile, err := os.Create(outputPrefix + ".jsonl")
+		if err != nil {
+			return "", err
+		}
+		defer jsonlFile.Close()
+		for _, file := range files {
+			if file.IsDir() || !strings.HasSuffix(file.Name(), ".txt") {
+				continue
+			}
+			txtFile, err := os.Open(input + "/" + file.Name())
+			if err != nil {
+				return "", err
+			}
+			defer txtFile.Close()
+			jsonlFile.WriteString("{\"text\": \"")
+			buf := make([]byte, 1024)
+			for {
+				n, err := txtFile.Read(buf)
+				if err != nil {
+					break
+				}
+				// regex replace \r\n \n \r with \\n
+				jsonlFile.WriteString(
+					strings.ReplaceAll(
+						strings.ReplaceAll(
+							strings.ReplaceAll(
+								strings.ReplaceAll(string(buf[:n]),
+									"\r\n", "\\n"),
+								"\n", "\\n"),
+							"\r", "\\n"),
+						"\n\n", "\\n"))
+			}
+			jsonlFile.WriteString("\"}\n")
+		}
+		input = outputPrefix + ".jsonl"
+	} else if err != nil {
+		return "", err
+	}
+
 	return Cmd(python, "./finetune/json2binidx_tool/tools/preprocess_data.py", "--input", input, "--output-prefix", outputPrefix, "--vocab", vocab,
 		"--tokenizer-type", tokenizerType, "--dataset-impl", "mmap", "--append-eod")
 }
@@ -113,3 +159,11 @@ func (a *App) InstallPyDep(python string, cnMirror bool) (string, error) {
 		return Cmd(python, "-m", "pip", "install", "-r", "./backend-python/requirements_without_cyac.txt")
 	}
 }
+
+func (a *App) GetPyError() string {
+	content, err := os.ReadFile("./error.txt")
+	if err != nil {
+		return ""
+	}
+	return string(content)
+}

backend-golang/wsl.go

@@ -119,8 +119,10 @@ func (a *App) WslStop() error {
 	if !running {
 		return errors.New("wsl not running")
 	}
-	err = cmd.Process.Kill()
-	cmd = nil
+	if cmd != nil {
+		err = cmd.Process.Kill()
+		cmd = nil
+	}
 	// stdin.Close()
 	stdin = nil
 	distro = nil

backend-python/convert_model.py

@@ -219,13 +219,17 @@ def get_args():
     return p.parse_args()
 
 
-args = get_args()
-if not args.quiet:
-    print(f"** {args}")
+try:
+    args = get_args()
+    if not args.quiet:
+        print(f"** {args}")
 
-RWKV(
-    getattr(args, "in"),
-    args.strategy,
-    verbose=not args.quiet,
-    convert_and_save_and_exit=args.out,
-)
+    RWKV(
+        getattr(args, "in"),
+        args.strategy,
+        verbose=not args.quiet,
+        convert_and_save_and_exit=args.out,
+    )
+except Exception as e:
+    with open("error.txt", "w") as f:
+        f.write(str(e))

finetune/install-wsl-dep-and-train.sh

@@ -8,6 +8,12 @@ if [[ ${cnMirror} == 1 ]]; then
   fi
 fi
 
+if dpkg -s "gcc" >/dev/null 2>&1; then
+  echo "gcc installed"
+else
+  sudo apt -y install gcc
+fi
+
 if dpkg -s "python3-pip" >/dev/null 2>&1; then
   echo "pip installed"
 else
@@ -20,14 +26,14 @@ else
   sudo apt -y install ninja-build
 fi
 
-if dpkg -s "cuda" >/dev/null 2>&1; then
-  echo "cuda installed"
+if dpkg -s "cuda" >/dev/null 2>&1 && dpkg -s "cuda" | grep Version | awk '{print $2}' | grep -q "12"; then
+  echo "cuda 12 installed"
 else
-  wget https://developer.download.nvidia.com/compute/cuda/repos/wsl-ubuntu/x86_64/cuda-wsl-ubuntu.pin
+  wget -N https://developer.download.nvidia.com/compute/cuda/repos/wsl-ubuntu/x86_64/cuda-wsl-ubuntu.pin
   sudo mv cuda-wsl-ubuntu.pin /etc/apt/preferences.d/cuda-repository-pin-600
-  wget https://developer.download.nvidia.com/compute/cuda/11.7.0/local_installers/cuda-repo-wsl-ubuntu-11-7-local_11.7.0-1_amd64.deb
-  sudo dpkg -i cuda-repo-wsl-ubuntu-11-7-local_11.7.0-1_amd64.deb
-  sudo cp /var/cuda-repo-wsl-ubuntu-11-7-local/cuda-*-keyring.gpg /usr/share/keyrings/
+  wget -N https://developer.download.nvidia.com/compute/cuda/12.2.0/local_installers/cuda-repo-wsl-ubuntu-12-2-local_12.2.0-1_amd64.deb
+  sudo dpkg -i cuda-repo-wsl-ubuntu-12-2-local_12.2.0-1_amd64.deb
+  sudo cp /var/cuda-repo-wsl-ubuntu-12-2-local/cuda-*-keyring.gpg /usr/share/keyrings/
   sudo apt-get update
   sudo apt-get -y install cuda
 fi

finetune/json2binidx_tool/tools/preprocess_data.py

@@ -17,11 +17,14 @@
 
 """Processing data for pretraining."""
 
-import argparse
-import multiprocessing
 import os
 import sys
 
+sys.path.append(os.path.dirname(os.path.realpath(__file__)))
+
+import argparse
+import multiprocessing
+
 import lm_dataformat as lmd
 import numpy as np
 
@@ -240,4 +243,8 @@ def main():
 
 
 if __name__ == "__main__":
-    main()
+    try:
+        main()
+    except Exception as e:
+        with open("error.txt", "w") as f:
+            f.write(str(e))

finetune/lora/merge_lora.py

@@ -5,49 +5,64 @@ from typing import Dict
 import typing
 import torch
 
-if '-h' in sys.argv or '--help' in sys.argv:
-    print(f'Usage: python3 {sys.argv[0]} [--use-gpu] <lora_alpha> <base_model.pth> <lora_checkpoint.pth> <output.pth>')
+try:
+    if "-h" in sys.argv or "--help" in sys.argv:
+        print(
+            f"Usage: python3 {sys.argv[0]} [--use-gpu] <lora_alpha> <base_model.pth> <lora_checkpoint.pth> <output.pth>"
+        )
 
-if sys.argv[1] == '--use-gpu':
-    device = 'cuda'
-    lora_alpha, base_model, lora, output = float(sys.argv[2]), sys.argv[3], sys.argv[4], sys.argv[5]
-else:
-    device = 'cpu'
-    lora_alpha, base_model, lora, output = float(sys.argv[1]), sys.argv[2], sys.argv[3], sys.argv[4]
+    if sys.argv[1] == "--use-gpu":
+        device = "cuda"
+        lora_alpha, base_model, lora, output = (
+            float(sys.argv[2]),
+            sys.argv[3],
+            sys.argv[4],
+            sys.argv[5],
+        )
+    else:
+        device = "cpu"
+        lora_alpha, base_model, lora, output = (
+            float(sys.argv[1]),
+            sys.argv[2],
+            sys.argv[3],
+            sys.argv[4],
+        )
 
+    with torch.no_grad():
+        w: Dict[str, torch.Tensor] = torch.load(base_model, map_location="cpu")
+        # merge LoRA-only slim checkpoint into the main weights
+        w_lora: Dict[str, torch.Tensor] = torch.load(lora, map_location="cpu")
+        for k in w_lora.keys():
+            w[k] = w_lora[k]
+        output_w: typing.OrderedDict[str, torch.Tensor] = OrderedDict()
+        # merge LoRA weights
+        keys = list(w.keys())
+        for k in keys:
+            if k.endswith(".weight"):
+                prefix = k[: -len(".weight")]
+                lora_A = prefix + ".lora_A"
+                lora_B = prefix + ".lora_B"
+                if lora_A in keys:
+                    assert lora_B in keys
+                    print(f"merging {lora_A} and {lora_B} into {k}")
+                    assert w[lora_B].shape[1] == w[lora_A].shape[0]
+                    lora_r = w[lora_B].shape[1]
+                    w[k] = w[k].to(device=device)
+                    w[lora_A] = w[lora_A].to(device=device)
+                    w[lora_B] = w[lora_B].to(device=device)
+                    w[k] += w[lora_B] @ w[lora_A] * (lora_alpha / lora_r)
+                    output_w[k] = w[k].to(device="cpu", copy=True)
+                    del w[k]
+                    del w[lora_A]
+                    del w[lora_B]
+                    continue
 
-with torch.no_grad():
-    w: Dict[str, torch.Tensor] = torch.load(base_model, map_location='cpu')
-    # merge LoRA-only slim checkpoint into the main weights
-    w_lora: Dict[str, torch.Tensor] = torch.load(lora, map_location='cpu')
-    for k in w_lora.keys():
-        w[k] = w_lora[k]
-    output_w: typing.OrderedDict[str, torch.Tensor] = OrderedDict()
-    # merge LoRA weights
-    keys = list(w.keys())
-    for k in keys:
-        if k.endswith('.weight'):
-            prefix = k[:-len('.weight')]
-            lora_A = prefix + '.lora_A'
-            lora_B = prefix + '.lora_B'
-            if lora_A in keys:
-                assert lora_B in keys
-                print(f'merging {lora_A} and {lora_B} into {k}')
-                assert w[lora_B].shape[1] == w[lora_A].shape[0]
-                lora_r = w[lora_B].shape[1]
-                w[k] = w[k].to(device=device)
-                w[lora_A] = w[lora_A].to(device=device)
-                w[lora_B] = w[lora_B].to(device=device)
-                w[k] += w[lora_B] @ w[lora_A] * (lora_alpha / lora_r)
-                output_w[k] = w[k].to(device='cpu', copy=True)
+            if "lora" not in k:
+                print(f"retaining {k}")
+                output_w[k] = w[k].clone()
                 del w[k]
-                del w[lora_A]
-                del w[lora_B]
-                continue
 
-        if 'lora' not in k:
-            print(f'retaining {k}')
-            output_w[k] = w[k].clone()
-            del w[k]
-
-    torch.save(output_w, output)
+        torch.save(output_w, output)
+except Exception as e:
+    with open("error.txt", "w") as f:
+        f.write(str(e))

frontend/src/_locales/zh-hans/main.json

@@ -223,5 +223,14 @@
   "Merge model successfully": "合并模型成功",
   "Convert Data successfully": "数据转换成功",
   "Please select a LoRA model": "请选择一个LoRA模型",
-  "You are using sample data for training. For formal training, please make sure to create your own jsonl file.": "你正在使用示例数据训练，对于正式训练场合，请务必创建你自己的jsonl训练数据"
+  "You are using sample data for training. For formal training, please make sure to create your own jsonl file.": "你正在使用示例数据训练，对于正式训练场合，请务必创建你自己的jsonl训练数据",
+  "WSL is not running. You may be using an outdated version of WSL, run \"wsl --update\" to update.": "WSL没有运行。你可能正在使用旧版本的WSL，请在cmd执行\"wsl --update\"以更新",
+  "Memory is not enough, try to increase the virtual memory or use a smaller base model.": "内存不足，尝试增加虚拟内存，或使用一个更小规模的基底模型",
+  "VRAM is not enough": "显存不足",
+  "Training data is not enough, reduce context length or add more data for training": "训练数据不足，请减小上下文长度或增加训练数据",
+  "You are using WSL 1 for training, please upgrade to WSL 2. e.g. Run \"wsl --set-version Ubuntu-22.04 2\"": "你正在使用WSL 1进行训练，请升级到WSL 2。例如，运行\"wsl --set-version Ubuntu-22.04 2\"",
+  "Matched CUDA is not installed": "未安装匹配的CUDA",
+  "Failed to convert data": "数据转换失败",
+  "Failed to merge model": "合并模型失败",
+  "The data path should be a directory or a file in jsonl format (more formats will be supported in the future).\n\nWhen you provide a directory path, all the txt files within that directory will be automatically converted into training data. This is commonly used for large-scale training in writing, code generation, or knowledge bases.\n\nThe jsonl format file can be referenced at https://github.com/Abel2076/json2binidx_tool/blob/main/sample.jsonl.\nYou can also write it similar to OpenAI's playground format, as shown in https://platform.openai.com/playground/p/default-chat.\nEven for multi-turn conversations, they must be written in a single line using `\\n` to indicate line breaks. If they are different dialogues or topics, they should be written in separate lines.": "数据路径必须是一个文件夹，或者jsonl格式文件 (未来会支持更多格式)\n\n当你填写的路径是一个文件夹时，该文件夹内的所有txt文件会被自动转换为训练数据，通常这用于大批量训练写作，代码生成或知识库\n\njsonl文件的格式参考 https://github.com/Abel2076/json2binidx_tool/blob/main/sample.jsonl\n你也可以仿照openai的playground编写，参考 https://platform.openai.com/playground/p/default-chat\n即使是多轮对话也必须写在一行，用`\\n`表示换行，如果是不同对话或主题，则另起一行"
 }

frontend/src/components/DialogButton.tsx

@@ -11,6 +11,7 @@ import {
 } from '@fluentui/react-components';
 import { ToolTipButton } from './ToolTipButton';
 import { useTranslation } from 'react-i18next';
+import MarkdownRender from './MarkdownRender';
 
 export const DialogButton: FC<{
   text?: string | null
@@ -19,12 +20,13 @@ export const DialogButton: FC<{
   className?: string,
   title: string,
   contentText: string,
-  onConfirm: () => void,
+  markdown?: boolean,
+  onConfirm?: () => void,
   size?: 'small' | 'medium' | 'large',
   shape?: 'rounded' | 'circular' | 'square',
   appearance?: 'secondary' | 'primary' | 'outline' | 'subtle' | 'transparent',
 }> = ({
-  text, icon, tooltip, className, title, contentText,
+  text, icon, tooltip, className, title, contentText, markdown,
   onConfirm, size, shape, appearance
 }) => {
   const { t } = useTranslation();
@@ -41,7 +43,11 @@ export const DialogButton: FC<{
       <DialogBody>
         <DialogTitle>{title}</DialogTitle>
         <DialogContent>
-          {contentText}
+          {
+            markdown ?
+              <MarkdownRender>{contentText}</MarkdownRender> :
+              contentText
+          }
         </DialogContent>
         <DialogActions>
           <DialogTrigger disableButtonEnhancement>

frontend/src/pages/Configs.tsx

@@ -13,8 +13,8 @@ import { Page } from '../components/Page';
 import { useNavigate } from 'react-router';
 import { RunButton } from '../components/RunButton';
 import { updateConfig } from '../apis';
-import { ConvertModel, FileExists } from '../../wailsjs/go/backend_golang/App';
-import { getStrategy, refreshLocalModels } from '../utils';
+import { ConvertModel, FileExists, GetPyError } from '../../wailsjs/go/backend_golang/App';
+import { getStrategy } from '../utils';
 import { useTranslation } from 'react-i18next';
 import { WindowShow } from '../../wailsjs/runtime/runtime';
 import strategyImg from '../assets/images/strategy.jpg';
@@ -253,9 +253,12 @@ export const Configs: FC = observer(() => {
                       const strategy = getStrategy(selectedConfig);
                       const newModelPath = modelPath + '-' + strategy.replace(/[:> *+]/g, '-');
                       toast(t('Start Converting'), { autoClose: 1000, type: 'info' });
-                      ConvertModel(commonStore.settings.customPythonPath, modelPath, strategy, newModelPath).then(() => {
-                        toast(`${t('Convert Success')} - ${newModelPath}`, { type: 'success' });
-                        refreshLocalModels({ models: commonStore.modelSourceList }, false);
+                      ConvertModel(commonStore.settings.customPythonPath, modelPath, strategy, newModelPath).then(async () => {
+                        if (!await FileExists(newModelPath)) {
+                          toast(t('Convert Failed') + ' - ' + await GetPyError(), { type: 'error' });
+                        } else {
+                          toast(`${t('Convert Success')} - ${newModelPath}`, { type: 'success' });
+                        }
                       }).catch(e => {
                         const errMsg = e.message || e;
                         if (errMsg.includes('path contains space'))

frontend/src/pages/Downloads.tsx

@@ -1,10 +1,10 @@
-import React, { FC, useEffect } from 'react';
+import React, { FC } from 'react';
 import { useTranslation } from 'react-i18next';
 import { Page } from '../components/Page';
 import { observer } from 'mobx-react-lite';
 import commonStore from '../stores/commonStore';
 import { Divider, Field, ProgressBar } from '@fluentui/react-components';
-import { bytesToGb, bytesToKb, bytesToMb, refreshLocalModels } from '../utils';
+import { bytesToGb, bytesToKb, bytesToMb } from '../utils';
 import { ToolTipButton } from '../components/ToolTipButton';
 import { Folder20Regular, Pause20Regular, Play20Regular } from '@fluentui/react-icons';
 import { AddToDownloadList, OpenFileFolder, PauseDownload } from '../../wailsjs/go/backend_golang/App';
@@ -23,12 +23,6 @@ export type DownloadStatus = {
 
 export const Downloads: FC = observer(() => {
   const { t } = useTranslation();
-  const finishedModelsLen = commonStore.downloadList.filter((status) => status.done && status.name.endsWith('.pth')).length;
-  useEffect(() => {
-    if (finishedModelsLen > 0)
-      refreshLocalModels({ models: commonStore.modelSourceList }, false);
-    console.log('finishedModelsLen:', finishedModelsLen);
-  }, [finishedModelsLen]);
 
   let displayList = commonStore.downloadList.slice();
   const downloadListNames = displayList.map(s => s.name);

frontend/src/pages/Train.tsx

@@ -4,6 +4,7 @@ import { Button, Dropdown, Input, Option, Select, Switch, Tab, TabList } from '@
 import {
   ConvertData,
   FileExists,
+  GetPyError,
   MergeLora,
   OpenFileFolder,
   WslCommand,
@@ -17,7 +18,7 @@ import { toast } from 'react-toastify';
 import commonStore from '../stores/commonStore';
 import { observer } from 'mobx-react-lite';
 import { SelectTabEventHandler } from '@fluentui/react-tabs';
-import { checkDependencies, refreshLocalModels, toastWithButton } from '../utils';
+import { checkDependencies, toastWithButton } from '../utils';
 import { Section } from '../components/Section';
 import { Labeled } from '../components/Labeled';
 import { ToolTipButton } from '../components/ToolTipButton';
@@ -37,6 +38,8 @@ import {
 import { Line } from 'react-chartjs-2';
 import { ChartJSOrUndefined } from 'react-chartjs-2/dist/types';
 import { WindowShow } from '../../wailsjs/runtime';
+import { t } from 'i18next';
+import { DialogButton } from '../components/DialogButton';
 
 ChartJS.register(
   CategoryScale,
@@ -49,15 +52,16 @@ ChartJS.register(
 );
 
 const parseLossData = (data: string) => {
-  const regex = /Epoch (\d+):\s+(\d+%)\|[\s\S]*\| (\d+)\/(\d+) \[(\d+:\d+)<(\d+:\d+),\s+(\d+.\d+it\/s), loss=(\d+.\d+),[\s\S]*\]/g;
+  const regex = /Epoch (\d+):\s+(\d+%)\|[\s\S]*\| (\d+)\/(\d+) \[(\d+:\d+)<(\d+:\d+),\s+(\d+.\d+it\/s), loss=(\S+),[\s\S]*\]/g;
   const matches = Array.from(data.matchAll(regex));
   if (matches.length === 0)
-    return;
+    return false;
   const lastMatch = matches[matches.length - 1];
   const epoch = parseInt(lastMatch[1]);
   const loss = parseFloat(lastMatch[8]);
   commonStore.setChartTitle(`Epoch ${epoch}: ${lastMatch[2]} - ${lastMatch[3]}/${lastMatch[4]} - ${lastMatch[5]}/${lastMatch[6]} - ${lastMatch[7]} Loss=${loss}`);
   addLossDataToChart(epoch, loss);
+  return true;
 };
 
 let chartLine: ChartJSOrUndefined<'line', (number | null)[], string>;
@@ -140,10 +144,36 @@ const loraFinetuneParametersOptions: Array<[key: keyof LoraFinetuneParameters, t
   ['headQk', 'boolean', 'Head QK']
 ];
 
+const showError = (e: any) => {
+  const msg = e.message || e;
+  if (msg === 'wsl not running') {
+    toast(t('WSL is not running. You may be using an outdated version of WSL, run "wsl --update" to update.'), { type: 'error' });
+  } else {
+    toast(t(msg), { type: 'error' });
+  }
+};
+
+const errorsMap = Object.entries({
+  'killed python3 ./finetune/lora/train.py': 'Memory is not enough, try to increase the virtual memory or use a smaller base model.',
+  'cuda out of memory': 'VRAM is not enough',
+  'valueerror: high <= 0': 'Training data is not enough, reduce context length or add more data for training',
+  '+= \'+ptx\'': 'You are using WSL 1 for training, please upgrade to WSL 2. e.g. Run "wsl --set-version Ubuntu-22.04 2"',
+  'cuda_home environment variable is not set': 'Matched CUDA is not installed',
+  'unsupported gpu architecture': 'Matched CUDA is not installed',
+  'error building extension \'fused_adam\'': 'Matched CUDA is not installed'
+});
+
 export const wslHandler = (data: string) => {
   if (data) {
     addWslMessage(data);
-    parseLossData(data);
+    const ok = parseLossData(data);
+    if (!ok)
+      for (const [key, value] of errorsMap) {
+        if (data.toLowerCase().includes(key)) {
+          showError(value);
+          return;
+        }
+      }
   }
 };
 
@@ -188,12 +218,8 @@ const Terminal: FC = observer(() => {
       WslStart().then(() => {
         addWslMessage('WSL> ' + input);
         setInput('');
-        WslCommand(input).catch((e: any) => {
-          toast((e.message || e), { type: 'error' });
-        });
-      }).catch((e: any) => {
-        toast((e.message || e), { type: 'error' });
-      });
+        WslCommand(input).catch(showError);
+      }).catch(showError);
     }
   };
 
@@ -208,9 +234,7 @@ const Terminal: FC = observer(() => {
         <Button onClick={() => {
           WslStop().then(() => {
             toast(t('Command Stopped'), { type: 'success' });
-          }).catch((e: any) => {
-            toast((e.message || e), { type: 'error' });
-          });
+          }).catch(showError);
         }}>
           {t('Stop')}
         </Button>
@@ -256,7 +280,7 @@ const LoraFinetune: FC = observer(() => {
     if (!ok)
       return;
 
-    const convertedDataPath = `./finetune/json2binidx_tool/data/${dataParams.dataPath.split('/').pop()!.split('.')[0]}_text_document`;
+    const convertedDataPath = `./finetune/json2binidx_tool/data/${dataParams.dataPath.split(/[\/\\]/).pop()!.split('.')[0]}_text_document`;
     if (!await FileExists(convertedDataPath + '.idx')) {
       toast(t('Please convert data first.'), { type: 'error' });
       return;
@@ -288,6 +312,7 @@ const LoraFinetune: FC = observer(() => {
         });
         WslCommand(`export cnMirror=${commonStore.settings.cnMirror ? '1' : '0'} ` +
           `&& export loadModel=models/${loraParams.baseModel} ` +
+          `&& sed -i 's/\\r$//' finetune/install-wsl-dep-and-train.sh ` +
           `&& chmod +x finetune/install-wsl-dep-and-train.sh && ./finetune/install-wsl-dep-and-train.sh ` +
           (loraParams.baseModel ? `--load_model models/${loraParams.baseModel} ` : '') +
           (loraParams.loraLoad ? `--lora_load lora-models/${loraParams.loraLoad} ` : '') +
@@ -301,9 +326,7 @@ const LoraFinetune: FC = observer(() => {
           `--beta1 ${loraParams.beta1} --beta2 ${loraParams.beta2} --adam_eps ${loraParams.adamEps} ` +
           `--devices ${loraParams.devices} --precision ${loraParams.precision} ` +
           `--grad_cp ${loraParams.gradCp ? '1' : '0'} ` +
-          `--lora_r ${loraParams.loraR} --lora_alpha ${loraParams.loraAlpha} --lora_dropout ${loraParams.loraDropout}`).catch((e: any) => {
-          toast((e.message || e), { type: 'error' });
-        });
+          `--lora_r ${loraParams.loraR} --lora_alpha ${loraParams.loraAlpha} --lora_dropout ${loraParams.loraDropout}`).catch(showError);
       }).catch(e => {
         const msg = e.message || e;
         if (msg === 'ubuntu not found') {
@@ -331,9 +354,7 @@ const LoraFinetune: FC = observer(() => {
               type: 'info',
               autoClose: false
             });
-          }).catch(e => {
-            toast((e.message || e), { type: 'error' });
-          });
+          }).catch(showError);
         });
       };
 
@@ -342,7 +363,7 @@ const LoraFinetune: FC = observer(() => {
       } else if (msg.includes('wsl.state: The system cannot find the file')) {
         enableWsl(true);
       } else {
-        toast(msg, { type: 'error' });
+        showError(msg);
       }
     });
   };
@@ -380,32 +401,43 @@ const LoraFinetune: FC = observer(() => {
           title={t('Data Process')}
           content={
             <div className="flex flex-col gap-2">
-              <Labeled flex label={t('Data Path')}
-                content={
-                  <div className="grow flex gap-2">
-                    <Input className="grow ml-2" value={dataParams.dataPath}
-                      onChange={(e, data) => {
-                        setDataParams({ dataPath: data.value });
-                      }} />
-                    <ToolTipButton desc={t('Open Folder')} icon={<Folder20Regular />} onClick={() => {
-                      OpenFileFolder(dataParams.dataPath, false);
-                    }} />
-                  </div>
-                } />
+              <div className="flex gap-2 items-center">
+                {t('Data Path')}
+                <Input className="grow" style={{ minWidth: 0 }} value={dataParams.dataPath}
+                  onChange={(e, data) => {
+                    setDataParams({ dataPath: data.value });
+                  }} />
+                <DialogButton text={t('Help')} title={t('Help')} markdown
+                  contentText={t('The data path should be a directory or a file in jsonl format (more formats will be supported in the future).\n\n' +
+                    'When you provide a directory path, all the txt files within that directory will be automatically converted into training data. ' +
+                    'This is commonly used for large-scale training in writing, code generation, or knowledge bases.\n\n' +
+                    'The jsonl format file can be referenced at https://github.com/Abel2076/json2binidx_tool/blob/main/sample.jsonl.\n' +
+                    'You can also write it similar to OpenAI\'s playground format, as shown in https://platform.openai.com/playground/p/default-chat.\n' +
+                    'Even for multi-turn conversations, they must be written in a single line using `\\n` to indicate line breaks. ' +
+                    'If they are different dialogues or topics, they should be written in separate lines.')} />
+                <ToolTipButton desc={t('Open Folder')} icon={<Folder20Regular />} onClick={() => {
+                  OpenFileFolder(dataParams.dataPath, false);
+                }} />
+              </div>
               <div className="flex gap-2 items-center">
                 {t('Vocab Path')}
                 <Input className="grow" style={{ minWidth: 0 }} value={dataParams.vocabPath}
                   onChange={(e, data) => {
                     setDataParams({ vocabPath: data.value });
                   }} />
-                <Button appearance="secondary" size="large" onClick={() => {
-                  ConvertData(commonStore.settings.customPythonPath, dataParams.dataPath,
-                    './finetune/json2binidx_tool/data/' + dataParams.dataPath.split('/').pop()!.split('.')[0],
-                    dataParams.vocabPath).then(() => {
-                    toast(t('Convert Data successfully'), { type: 'success' });
-                  }).catch((e: any) => {
-                    toast((e.message || e), { type: 'error' });
-                  });
+                <Button appearance="secondary" onClick={async () => {
+                  const ok = await checkDependencies(navigate);
+                  if (!ok)
+                    return;
+                  const outputPrefix = './finetune/json2binidx_tool/data/' +
+                    dataParams.dataPath.replace(/[\/\\]$/, '').split(/[\/\\]/).pop()!.split('.')[0];
+                  ConvertData(commonStore.settings.customPythonPath, dataParams.dataPath, outputPrefix, dataParams.vocabPath).then(async () => {
+                    if (!await FileExists(outputPrefix + '_text_document.idx')) {
+                      toast(t('Failed to convert data') + ' - ' + await GetPyError(), { type: 'error' });
+                    } else {
+                      toast(t('Convert Data successfully'), { type: 'success' });
+                    }
+                  }).catch(showError);
                 }}>{t('Convert')}</Button>
               </div>
             </div>
@@ -452,14 +484,16 @@ const LoraFinetune: FC = observer(() => {
                 if (!ok)
                   return;
                 if (loraParams.loraLoad) {
+                  const outputPath = `models/${loraParams.baseModel}-LoRA-${loraParams.loraLoad}`;
                   MergeLora(commonStore.settings.customPythonPath, true, loraParams.loraAlpha,
                     'models/' + loraParams.baseModel, 'lora-models/' + loraParams.loraLoad,
-                    `models/${loraParams.baseModel}-LoRA-${loraParams.loraLoad}`).then(() => {
-                    toast(t('Merge model successfully'), { type: 'success' });
-                    refreshLocalModels({ models: commonStore.modelSourceList }, false);
-                  }).catch((e: any) => {
-                    toast((e.message || e), { type: 'error' });
-                  });
+                    outputPath).then(async () => {
+                    if (!await FileExists(outputPath)) {
+                      toast(t('Failed to merge model') + ' - ' + await GetPyError(), { type: 'error' });
+                    } else {
+                      toast(t('Merge model successfully'), { type: 'success' });
+                    }
+                  }).catch(showError);
                 } else {
                   toast(t('Please select a LoRA model'), { type: 'info' });
                 }
@@ -521,9 +555,7 @@ const LoraFinetune: FC = observer(() => {
         <Button appearance="secondary" size="large" onClick={() => {
           WslStop().then(() => {
             toast(t('Command Stopped'), { type: 'success' });
-          }).catch((e: any) => {
-            toast((e.message || e), { type: 'error' });
-          });
+          }).catch(showError);
         }}>{t('Stop')}</Button>
         <Button appearance="primary" size="large" onClick={StartLoraFinetune}>{t('Train')}</Button>
       </div>

frontend/src/startup.ts

@@ -1,6 +1,6 @@
 import commonStore, { Platform } from './stores/commonStore';
 import { GetPlatform, ListDirFiles, ReadJson } from '../wailsjs/go/backend_golang/App';
-import { Cache, checkUpdate, downloadProgramFiles, LocalConfig, refreshModels } from './utils';
+import { Cache, checkUpdate, downloadProgramFiles, LocalConfig, refreshLocalModels, refreshModels } from './utils';
 import { getStatus } from './apis';
 import { EventsOn } from '../wailsjs/runtime';
 import manifest from '../../manifest.json';
@@ -18,6 +18,7 @@ export async function startup() {
   EventsOn('wslerr', (e) => {
     console.log(e);
   });
+  initLocalModelsNotify();
   initLoraModels();
 
   initPresets();
@@ -109,3 +110,10 @@ async function initLoraModels() {
       refreshLoraModels();
   });
 }
+
+async function initLocalModelsNotify() {
+  EventsOn('fsnotify', (data: string) => {
+    if (data.includes('models') && !data.includes('lora-models'))
+      refreshLocalModels({ models: commonStore.modelSourceList }, false); //TODO fix bug that only add models
+  });
+}

frontend/wailsjs/go/backend_golang/App.d.ts

@@ -22,6 +22,8 @@ export function FileExists(arg1:string):Promise<boolean>;
 
 export function GetPlatform():Promise<string>;
 
+export function GetPyError():Promise<string>;
+
 export function InstallPyDep(arg1:string,arg2:boolean):Promise<string>;
 
 export function ListDirFiles(arg1:string):Promise<Array<backend_golang.FileInfo>>;

frontend/wailsjs/go/backend_golang/App.js

@@ -42,6 +42,10 @@ export function GetPlatform() {
   return window['go']['backend_golang']['App']['GetPlatform']();
 }
 
+export function GetPyError() {
+  return window['go']['backend_golang']['App']['GetPyError']();
+}
+
 export function InstallPyDep(arg1, arg2) {
   return window['go']['backend_golang']['App']['InstallPyDep'](arg1, arg2);
 }

main.go

@@ -37,6 +37,7 @@ func main() {
 		backend.CopyEmbed(finetune)
 		os.Mkdir("models", os.ModePerm)
 		os.Mkdir("lora-models", os.ModePerm)
+		os.Mkdir("finetune/json2binidx_tool/data", os.ModePerm)
 	}
 
 	f, err := os.Create("lora-models/train_log.txt")

manifest.json

@@ -1,12 +1,12 @@
 {
-  "version": "1.3.2",
+  "version": "1.3.4",
   "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）。"
   },
   "about": {
-    "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>"
+    "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-World: https://huggingface.co/BlinkDL/rwkv-4-world/tree/main\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\nRWKV-LM-LoRA: https://github.com/Blealtan/RWKV-LM-LoRA\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-World: https://huggingface.co/BlinkDL/rwkv-4-world/tree/main\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\nRWKV-LM-LoRA: https://github.com/Blealtan/RWKV-LM-LoRA\n\n</div>"
   },
   "programFiles": [
     {