add support for dynamic state-tuned models

backend-golang/app.go

@@ -125,6 +125,7 @@ func (a *App) OnStartup(ctx context.Context) {
 	os.Chmod(a.exDir+"backend-rust/web-rwkv-converter", 0777)
 	os.Mkdir(a.exDir+"models", os.ModePerm)
 	os.Mkdir(a.exDir+"lora-models", os.ModePerm)
+	os.Mkdir(a.exDir+"state-models", os.ModePerm)
 	os.Mkdir(a.exDir+"finetune/json2binidx_tool/data", os.ModePerm)
 	trainLogPath := "lora-models/train_log.txt"
 	if !a.FileExists(trainLogPath) {
@@ -151,8 +152,9 @@ func (a *App) OnBeforeClose(ctx context.Context) bool {
 func (a *App) watchFs() {
 	watcher, err := fsnotify.NewWatcher()
 	if err == nil {
-		watcher.Add(a.exDir + "./lora-models")
 		watcher.Add(a.exDir + "./models")
+		watcher.Add(a.exDir + "./lora-models")
+		watcher.Add(a.exDir + "./state-models")
 		go func() {
 			for {
 				select {

backend-python/routes/config.py

@@ -120,6 +120,9 @@ def update_config(body: ModelConfigBody):
         model_config = ModelConfigBody()
         global_var.set(global_var.Model_Config, model_config)
     merge_model(model_config, body)
+    exception = load_rwkv_state(global_var.get(global_var.Model), model_config.state)
+    if exception is not None:
+        raise exception
     print("Updated Model Config:", model_config)
 
     return "success"

backend-python/routes/state_cache.py

@@ -176,6 +176,19 @@ def reset_state():
     return "success"
 
 
+def force_reset_state():
+    global trie, dtrie
+
+    if trie is None:
+        return
+
+    import cyac
+
+    trie = cyac.Trie()
+    dtrie = {}
+    gc.collect()
+
+
 class LongestPrefixStateBody(BaseModel):
     prompt: str
 

backend-python/utils/rwkv.py

@@ -7,7 +7,7 @@
 import time
 from typing import Dict, Iterable, List, Tuple, Union, Type, Callable
 from utils.log import quick_log
-from fastapi import HTTPException
+from fastapi import HTTPException, status
 from pydantic import BaseModel, Field
 from routes import state_cache
 import global_var
@@ -27,6 +27,7 @@ def __init__(self, model, pipeline):
         self.EOS_ID = 0
 
         self.name = "rwkv"
+        self.model_path = ""
         self.version = 4
         self.model = model
         self.pipeline = pipeline
@@ -43,6 +44,8 @@ def __init__(self, model, pipeline):
         self.penalty_alpha_frequency = 1
         self.penalty_decay = 0.996
         self.global_penalty = False
+        self.state_path = ""
+        self.state_tuned = None
 
     @abstractmethod
     def adjust_occurrence(self, occurrence: Dict, token: int):
@@ -236,7 +239,10 @@ def generate(
         except HTTPException:
             pass
         if cache is None or cache["prompt"] == "" or cache["state"] is None:
-            self.model_state = None
+            if self.state_path:
+                self.model_state = copy.deepcopy(self.state_tuned)
+            else:
+                self.model_state = None
             self.model_tokens = []
         else:
             delta_prompt = prompt[len(cache["prompt"]) :]
@@ -606,13 +612,13 @@ def get_model_path(model_path: str) -> str:
 
 
 def RWKV(model: str, strategy: str, tokenizer: Union[str, None]) -> AbstractRWKV:
-    model = get_model_path(model)
+    model_path = get_model_path(model)
 
     rwkv_beta = global_var.get(global_var.Args).rwkv_beta
     rwkv_cpp = getattr(global_var.get(global_var.Args), "rwkv.cpp")
     webgpu = global_var.get(global_var.Args).webgpu
 
-    if "midi" in model.lower() or "abc" in model.lower():
+    if "midi" in model_path.lower() or "abc" in model_path.lower():
         os.environ["RWKV_RESCALE_LAYER"] = "999"
 
     # dynamic import to make RWKV_CUDA_ON work
@@ -637,8 +643,8 @@ def RWKV(model: str, strategy: str, tokenizer: Union[str, None]) -> AbstractRWKV
         )
     from rwkv_pip.utils import PIPELINE
 
-    filename, _ = os.path.splitext(os.path.basename(model))
-    model = Model(model, strategy)
+    filename, _ = os.path.splitext(os.path.basename(model_path))
+    model = Model(model_path, strategy)
     if not tokenizer:
         tokenizer = get_tokenizer(len(model.w["emb.weight"]))
     pipeline = PIPELINE(model, tokenizer)
@@ -671,6 +677,7 @@ def RWKV(model: str, strategy: str, tokenizer: Union[str, None]) -> AbstractRWKV
         else:
             rwkv = TextRWKV(model, pipeline)
     rwkv.name = filename
+    rwkv.model_path = model_path
     rwkv.version = model.version
 
     return rwkv
@@ -688,6 +695,7 @@ class ModelConfigBody(BaseModel):
         default=None,
         description="When generating a response, whether to include the submitted prompt as a penalty factor. By turning this off, you will get the same generated results as official RWKV Gradio. If you find duplicate results in the generated results, turning this on can help avoid generating duplicates.",
     )
+    state: str = Field(default=None, description="state-tuned file path")
 
     model_config = {
         "json_schema_extra": {
@@ -699,11 +707,80 @@ class ModelConfigBody(BaseModel):
                 "frequency_penalty": 1,
                 "penalty_decay": 0.996,
                 "global_penalty": False,
+                "state": "",
             }
         }
     }
 
 
+def load_rwkv_state(model: AbstractRWKV, state_path: str) -> HTTPException:
+    if model:
+        if state_path:
+            if model.model_path.endswith(".pth") and state_path.endswith(".pth"):
+                import torch
+
+                state_path = get_model_path(state_path)
+                if model.state_path == state_path:
+                    return
+
+                state_raw = torch.load(state_path, map_location="cpu")
+                state_raw_shape = next(iter(state_raw.values())).shape
+
+                args = model.model.args
+                if (
+                    len(state_raw) != args.n_layer
+                    or state_raw_shape[0] * state_raw_shape[1] != args.n_embd
+                ):
+                    if model.state_path:
+                        pass
+                    else:
+                        print("state failed to load")
+                    return HTTPException(
+                        status.HTTP_400_BAD_REQUEST, "state shape mismatch"
+                    )
+
+                strategy = model.model.strategy
+                model.state_tuned = [None] * args.n_layer * 3
+
+                for i in range(args.n_layer):
+                    dd = strategy[i]
+                    dev = dd.device
+                    atype = dd.atype
+                    model.state_tuned[i * 3 + 0] = torch.zeros(
+                        args.n_embd, dtype=atype, requires_grad=False, device=dev
+                    ).contiguous()
+                    model.state_tuned[i * 3 + 1] = (
+                        state_raw[f"blocks.{i}.att.time_state"]
+                        .transpose(1, 2)
+                        .to(dtype=torch.float, device=dev)
+                        .requires_grad_(False)
+                        .contiguous()
+                    )
+                    model.state_tuned[i * 3 + 2] = torch.zeros(
+                        args.n_embd, dtype=atype, requires_grad=False, device=dev
+                    ).contiguous()
+
+                state_cache.force_reset_state()
+                model.state_path = state_path
+                print("state loaded")
+            else:
+                if model.state_path:
+                    pass
+                else:
+                    print("state failed to load")
+                return HTTPException(
+                    status.HTTP_400_BAD_REQUEST,
+                    "file format of the model or state model not supported",
+                )
+        else:
+            state_cache.force_reset_state()
+            model.state_path = ""
+            model.state_tuned = None  # TODO cached
+            print("state unloaded")
+    else:
+        print("state not loaded")
+
+
 def set_rwkv_config(model: AbstractRWKV, body: ModelConfigBody):
     if body.max_tokens is not None:
         model.max_tokens_per_generation = body.max_tokens
@@ -724,6 +801,8 @@ def set_rwkv_config(model: AbstractRWKV, body: ModelConfigBody):
         model.top_k = body.top_k
     if body.global_penalty is not None:
         model.global_penalty = body.global_penalty
+    if body.state is not None:
+        load_rwkv_state(model, body.state)
 
 
 def get_rwkv_config(model: AbstractRWKV) -> ModelConfigBody:
@@ -736,4 +815,5 @@ def get_rwkv_config(model: AbstractRWKV) -> ModelConfigBody:
         penalty_decay=model.penalty_decay,
         top_k=model.top_k,
         global_penalty=model.global_penalty,
+        state=model.state_path,
     )

frontend/src/_locales/ja/main.json

@@ -354,5 +354,10 @@
   "Inside the model, there is a default prompt to improve the model's handling of common issues, but it may degrade the role-playing effect. You can disable this option to achieve a better role-playing effect.": "モデル内部には、一般的な問題の処理を改善するためのデフォルトのプロンプトがありますが、役割演技の効果を低下させる可能性があります。このオプションを無効にすることで、より良い役割演技効果を得ることができます。",
   "Exit without saving": "保存せずに終了",
   "Content has been changed, are you sure you want to exit without saving?": "コンテンツが変更されています、保存せずに終了してもよろしいですか？",
-  "Don't forget to correctly fill in your Ollama API Chat Model Name.": "Ollama APIチャットモデル名を正しく記入するのを忘れないでください。"
+  "Don't forget to correctly fill in your Ollama API Chat Model Name.": "Ollama APIチャットモデル名を正しく記入するのを忘れないでください。",
+  "State-tuned Model": "State調整モデル",
+  "See More": "もっと見る",
+  "State Model": "Stateモデル",
+  "State model mismatch": "Stateモデルの不一致",
+  "File format of the model or state model not supported": "モデルまたはStateモデルのファイル形式がサポートされていません"
 }

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

@@ -354,5 +354,10 @@
   "Inside the model, there is a default prompt to improve the model's handling of common issues, but it may degrade the role-playing effect. You can disable this option to achieve a better role-playing effect.": "模型内部有一个默认提示来改善模型处理常规问题的效果, 但它可能会让角色扮演的效果变差, 你可以关闭此选项来获得更好的角色扮演效果",
   "Exit without saving": "退出而不保存",
   "Content has been changed, are you sure you want to exit without saving?": "内容已经被修改, 你确定要退出而不保存吗?",
-  "Don't forget to correctly fill in your Ollama API Chat Model Name.": "不要忘记正确填写你的Ollama API 聊天模型名"
+  "Don't forget to correctly fill in your Ollama API Chat Model Name.": "不要忘记正确填写你的Ollama API 聊天模型名",
+  "State-tuned Model": "State微调模型",
+  "See More": "查看更多",
+  "State Model": "State模型",
+  "State model mismatch": "State模型不匹配",
+  "File format of the model or state model not supported": "模型或state模型的文件格式不支持"
 }

frontend/src/components/RunButton.tsx

@@ -214,7 +214,18 @@ export const RunButton: FC<{ onClickRun?: MouseEventHandler, iconMode?: boolean
                 presence_penalty: modelConfig.apiParameters.presencePenalty,
                 frequency_penalty: modelConfig.apiParameters.frequencyPenalty,
                 penalty_decay: modelConfig.apiParameters.penaltyDecay,
-                global_penalty: modelConfig.apiParameters.globalPenalty
+                global_penalty: modelConfig.apiParameters.globalPenalty,
+                state: modelConfig.apiParameters.stateModel
+              }).then(async r => {
+                if (r.status !== 200) {
+                  const error = await r.text();
+                  if (error.includes('state shape mismatch'))
+                    toast(t('State model mismatch'), { type: 'error' });
+                  else if (error.includes('file format of the model or state model not supported'))
+                    toast(t('File format of the model or state model not supported'), { type: 'error' });
+                  else
+                    toast(error, { type: 'error' });
+                }
               });
             }
 

frontend/src/pages/Configs.tsx

@@ -7,11 +7,13 @@ import {
   Dropdown,
   Input,
   Label,
+  Link,
   Option,
   PresenceBadge,
   Select,
   Switch,
-  Text
+  Text,
+  Tooltip
 } from '@fluentui/react-components';
 import { AddCircle20Regular, DataUsageSettings20Regular, Delete20Regular, Save20Regular } from '@fluentui/react-icons';
 import React, { FC, useCallback, useEffect, useRef } from 'react';
@@ -27,7 +29,7 @@ import { Page } from '../components/Page';
 import { useNavigate } from 'react-router';
 import { RunButton } from '../components/RunButton';
 import { updateConfig } from '../apis';
-import { getStrategy } from '../utils';
+import { getStrategy, isDynamicStateSupported } from '../utils';
 import { useTranslation } from 'react-i18next';
 import strategyImg from '../assets/images/strategy.jpg';
 import strategyZhImg from '../assets/images/strategy_zh.jpg';
@@ -36,6 +38,7 @@ import { useMediaQuery } from 'usehooks-ts';
 import { ApiParameters, Device, ModelParameters, Precision } from '../types/configs';
 import { convertModel, convertToGGML, convertToSt } from '../utils/convert-model';
 import { defaultPenaltyDecay } from './defaultConfigs';
+import { BrowserOpenURL } from '../../wailsjs/runtime';
 
 const ConfigSelector: FC<{
   selectedIndex: number,
@@ -112,14 +115,27 @@ const Configs: FC = observer(() => {
 
   const onClickSave = () => {
     commonStore.setModelConfig(selectedIndex, selectedConfig);
+    // When clicking RunButton in Configs page, updateConfig will be called twice,
+    // because there are also RunButton in other pages, and the calls to updateConfig in both places are necessary.
     updateConfig({
       max_tokens: selectedConfig.apiParameters.maxResponseToken,
       temperature: selectedConfig.apiParameters.temperature,
       top_p: selectedConfig.apiParameters.topP,
       presence_penalty: selectedConfig.apiParameters.presencePenalty,
       frequency_penalty: selectedConfig.apiParameters.frequencyPenalty,
       penalty_decay: selectedConfig.apiParameters.penaltyDecay,
-      global_penalty: selectedConfig.apiParameters.globalPenalty
+      global_penalty: selectedConfig.apiParameters.globalPenalty,
+      state: selectedConfig.apiParameters.stateModel
+    }).then(async r => {
+      if (r.status !== 200) {
+        const error = await r.text();
+        if (error.includes('state shape mismatch'))
+          toast(t('State model mismatch'), { type: 'error' });
+        else if (error.includes('file format of the model or state model not supported'))
+          toast(t('File format of the model or state model not supported'), { type: 'error' });
+        else
+          toast(error, { type: 'error' });
+      }
     });
     toast(t('Config Saved'), { autoClose: 300, type: 'success' });
   };
@@ -200,6 +216,34 @@ const Configs: FC = observer(() => {
                         });
                       }} />
                   } />
+                {isDynamicStateSupported(selectedConfig) &&
+                  <div className="sm:col-span-2 flex gap-2 items-center min-w-0">
+                    <Tooltip content={<div>
+                      {t('State-tuned Model')}, {t('See More')}: <Link
+                      onClick={() => BrowserOpenURL('https://github.com/BlinkDL/RWKV-LM#state-tuning-tuning-the-initial-state-zero-inference-overhead')}>{'https://github.com/BlinkDL/RWKV-LM#state-tuning-tuning-the-initial-state-zero-inference-overhead'}
+                    </Link>
+                    </div>} showDelay={0} hideDelay={0}
+                      relationship="description">
+                      <div className="shrink-0">
+                        {t('State Model') + ' *'}
+                      </div>
+                    </Tooltip>
+                    <Select style={{ minWidth: 0 }} className="grow"
+                      value={selectedConfig.apiParameters.stateModel}
+                      onChange={(e, data) => {
+                        setSelectedConfigApiParams({
+                          stateModel: data.value
+                        });
+                      }}>
+                      <option key={-1} value={''}>
+                        {t('None')}
+                      </option>
+                      {commonStore.stateModels.map((modelName, index) =>
+                        <option key={index} value={modelName}>{modelName}</option>
+                      )}
+                    </Select>
+                  </div>
+                }
                 <Accordion className="sm:col-span-2" collapsible
                   openItems={!commonStore.apiParamsCollapsed && 'advanced'}
                   onToggle={(e, data) => {