Merge pull request #714 from Ikaros-521/owner

千帆大模型支持千帆AppBuilder，可以接入自定义应用（对接请更新WenxinWorkshop
Ikaros-521 · Mar 21, 2024 · 3e78c96 · 3e78c96
2 parents 9e7d578 + db09205
commit 3e78c96
Show file tree

Hide file tree

Showing 8 changed files with 131 additions and 70 deletions.
diff --git a/config.json b/config.json
@@ -351,9 +351,11 @@
     "history_max_len": 300
   },
   "my_wenxinworkshop": {
+    "type": "千帆大模型",
     "model": "ERNIEBot",
     "api_key": "",
     "secret_key": "",
+    "app_token": "",
     "top_p": 0.8,
     "temperature": 0.9,
     "penalty_score": 1.0,

diff --git a/config.json.bak b/config.json.bak
@@ -351,9 +351,11 @@
     "history_max_len": 300
   },
   "my_wenxinworkshop": {
+    "type": "千帆大模型",
     "model": "ERNIEBot",
     "api_key": "",
     "secret_key": "",
+    "app_token": "",
     "top_p": 0.8,
     "temperature": 0.9,
     "penalty_score": 1.0,

diff --git a/docs/投资人/index.html b/docs/投资人/index.html
@@ -286,6 +286,11 @@
             avatar: "https://images.cnblogs.com/cnblogs_com/ikaros-521/2328032/o_240219073301_QQ%E5%9B%BE%E7%89%8720240219153230.jpg",
             amount: "￥100"
         },
+        {
+            name: "不醉不归",
+            avatar: "https://images.cnblogs.com/cnblogs_com/ikaros-521/2328032/o_240320155329_QQ%E5%9B%BE%E7%89%8720240320235305.jpg",
+            amount: "￥100"
+        },
         {
             name: "很养眼的丑男",
             avatar: "https://images.cnblogs.com/cnblogs_com/ikaros-521/2328032/o_230722162732_QQ%E5%9B%BE%E7%89%8720230723000544.jpg",

diff --git a/docs/投资人/invest.png b/docs/投资人/invest.png
diff --git a/tests/test_blivedm/2.py b/tests/test_blivedm/2.py
@@ -5,13 +5,20 @@
 import blivedm.models.open_live as open_models
 import blivedm.models.web as web_models
 
+config_json = {
+    "ACCESS_KEY_ID": "",
+    "ACCESS_KEY_SECRET": "",
+    "APP_ID": 0,
+    "ROOM_OWNER_AUTH_CODE": ""
+}
+
 # 在开放平台申请的开发者密钥
-ACCESS_KEY_ID = ''
-ACCESS_KEY_SECRET = ''
+ACCESS_KEY_ID = config_json["ACCESS_KEY_ID"]
+ACCESS_KEY_SECRET = config_json["ACCESS_KEY_SECRET"]
 # 在开放平台创建的项目ID
-APP_ID = 0
+APP_ID = config_json["APP_ID"]
 # 主播身份码
-ROOM_OWNER_AUTH_CODE = ''
+ROOM_OWNER_AUTH_CODE = config_json["ROOM_OWNER_AUTH_CODE"]
 
 
 async def main():

diff --git a/utils/gpt_model/my_wenxinworkshop.py b/utils/gpt_model/my_wenxinworkshop.py
@@ -1,26 +1,60 @@
 import json, logging, traceback
-from wenxinworkshop import LLMAPI, EmbeddingAPI, PromptTemplateAPI
+from wenxinworkshop import LLMAPI, AppBuilderAPI, EmbeddingAPI, PromptTemplateAPI
 from wenxinworkshop import Message, Messages, Texts
 
+from utils.common import Common
+from utils.logger import Configure_logger
+
 # 前往官网：https://cloud.baidu.com/product/wenxinworkshop 申请服务获取
 
 class My_WenXinWorkShop:
     def __init__(self, data):
-        # self.common = Common()
-        # # 日志文件路径
-        # file_path = "./log/log-" + self.common.get_bj_time(1) + ".txt"
-        # Configure_logger(file_path)
+        self.common = Common()
+        # 日志文件路径
+        file_path = "./log/log-" + self.common.get_bj_time(1) + ".txt"
+        Configure_logger(file_path)
 
         self.config_data = data
         self.history = []
 
+        self.my_bot = None
+
+        logging.debug(self.config_data)
+
         try:
-            # create a LLM API
-            self.my_bot = LLMAPI(
-                api_key=self.config_data["api_key"],
-                secret_key=self.config_data["secret_key"],
-                url=LLMAPI.ERNIEBot
-            )
+            if self.config_data['type'] == "千帆大模型":
+                model_url_map = {
+                    "ERNIEBot": LLMAPI.ERNIEBot,
+                    "ERNIEBot_turbo": LLMAPI.ERNIEBot_turbo,
+                    "ERNIEBot_4_0": LLMAPI.ERNIEBot_4_0,
+                    "BLOOMZ_7B": LLMAPI.BLOOMZ_7B,
+                    "LLAMA_2_7B": LLMAPI.LLAMA_2_7B,
+                    "LLAMA_2_13B": LLMAPI.LLAMA_2_13B,
+                    "LLAMA_2_70B": LLMAPI.LLAMA_2_70B,
+                    "ERNIEBot_4_0": LLMAPI.ERNIEBot_4_0,
+                    "QIANFAN_BLOOMZ_7B_COMPRESSED": LLMAPI.QIANFAN_BLOOMZ_7B_COMPRESSED,
+                    "QIANFAN_CHINESE_LLAMA_2_7B": LLMAPI.QIANFAN_CHINESE_LLAMA_2_7B,
+                    "CHATGLM2_6B_32K": LLMAPI.CHATGLM2_6B_32K,
+                    "AQUILACHAT_7B": LLMAPI.AQUILACHAT_7B,
+                    "ERNIE_BOT_8K": LLMAPI.ERNIE_BOT_8K,
+                    "CODELLAMA_7B_INSTRUCT": LLMAPI.CODELLAMA_7B_INSTRUCT,
+                    "XUANYUAN_70B_CHAT": LLMAPI.XUANYUAN_70B_CHAT,
+                    "CHATLAW": LLMAPI.QIANFAN_BLOOMZ_7B_COMPRESSED,
+                    "QIANFAN_BLOOMZ_7B_COMPRESSED": LLMAPI.CHATLAW,
+                }
+
+                selected_model = self.config_data["model"]
+                if selected_model in model_url_map:
+                    self.my_bot = LLMAPI(
+                        api_key=self.config_data["api_key"],
+                        secret_key=self.config_data["secret_key"],
+                        url=model_url_map[selected_model]
+                    )
+            elif self.config_data['type'] == "AppBuilder":
+                self.my_bot = AppBuilderAPI(
+                    app_token=self.config_data["app_token"],
+                    history_enable=self.config_data["history_enable"]
+                )
         except Exception as e:
             logging.error(traceback.format_exc())
 
@@ -36,48 +70,54 @@ def get_resp(self, prompt):
             str: 返回的文本回答
         """
         try:
-            # create messages
-            messages: Messages = []
-
-            for history in self.history:
+            if self.config_data['type'] == "千帆大模型":
+                # create messages
+                messages: Messages = []
+
+                for history in self.history:
+                    messages.append(Message(
+                        role=history["role"],
+                        content=history["content"]
+                    ))
+
                 messages.append(Message(
-                    role=history["role"],
-                    content=history["content"]
+                    role='user',
+                    content=prompt
                 ))
 
-            messages.append(Message(
-                role='user',
-                content=prompt
-            ))
-
-            logging.info(f"self.history={self.history}")
-
-            # get response from LLM API
-            resp_content = self.my_bot(
-                messages=messages,
-                temperature=self.config_data["temperature"],
-                top_p=self.config_data["top_p"],
-                penalty_score=self.config_data["penalty_score"],
-                stream=None,
-                user_id=None,
-                chunk_size=512
-            )
-
-            # 启用历史就给我记住！
-            if self.config_data["history_enable"]:
-                while True:
-                    # 获取嵌套列表中所有字符串的字符数
-                    total_chars = sum(len(item['content']) for item in self.history if 'content' in item)
-                    # 如果大于限定最大历史数，就剔除第一个元素
-                    if total_chars > self.config_data["history_max_len"]:
-                        self.history.pop(0)
-                        self.history.pop(0)
-                    else:
-                        # self.history.pop()
-                        self.history.append({"role": "user", "content": prompt})
-                        self.history.append({"role": "assistant", "content": resp_content})
-                        break
-
+                logging.info(f"self.history={self.history}")
+
+                # get response from LLM API
+                resp_content = self.my_bot(
+                    messages=messages,
+                    temperature=self.config_data["temperature"],
+                    top_p=self.config_data["top_p"],
+                    penalty_score=self.config_data["penalty_score"],
+                    stream=None,
+                    user_id=None,
+                    chunk_size=512
+                )
+
+                # 启用历史就给我记住！
+                if self.config_data["history_enable"]:
+                    while True:
+                        # 获取嵌套列表中所有字符串的字符数
+                        total_chars = sum(len(item['content']) for item in self.history if 'content' in item)
+                        # 如果大于限定最大历史数，就剔除第一个元素
+                        if total_chars > self.config_data["history_max_len"]:
+                            self.history.pop(0)
+                            self.history.pop(0)
+                        else:
+                            # self.history.pop()
+                            self.history.append({"role": "user", "content": prompt})
+                            self.history.append({"role": "assistant", "content": resp_content})
+                            break
+            elif self.config_data['type'] == "AppBuilder":
+                resp_content = self.my_bot(
+                    query=prompt,
+                    response_mode="blocking"
+                )
+                
             return resp_content
 
         except Exception as e:

diff --git a/utils/my_handle.py b/utils/my_handle.py
@@ -1224,7 +1224,8 @@ def llm_handle(self, chat_type, data, type="chat"):
             # 使用字典映射的方式来获取响应内容
             resp_content = chat_model_methods.get(chat_type, lambda: data["content"])()
 
-            resp_content = resp_content.strip()
+            if resp_content is not None:
+                resp_content = resp_content.strip()
 
             logging.debug(f"resp_content={resp_content}")
 

diff --git a/webui.py b/webui.py
@@ -1432,6 +1432,7 @@ def common_textarea_handle(content):
                 # config_data["my_qianfan"]["history_max_len"] = int(input_my_qianfan_history_max_len.value)
 
                 if config.get("webui", "show_card", "llm", "my_wenxinworkshop"):
+                    config_data["my_wenxinworkshop"]["type"] = select_my_wenxinworkshop_type.value
                     config_data["my_wenxinworkshop"]["model"] = select_my_wenxinworkshop_model.value
                     config_data["my_wenxinworkshop"]["api_key"] = input_my_wenxinworkshop_api_key.value
                     config_data["my_wenxinworkshop"]["secret_key"] = input_my_wenxinworkshop_secret_key.value
@@ -1441,6 +1442,8 @@ def common_textarea_handle(content):
                     config_data["my_wenxinworkshop"]["history_enable"] = switch_my_wenxinworkshop_history_enable.value
                     config_data["my_wenxinworkshop"]["history_max_len"] = int(input_my_wenxinworkshop_history_max_len.value)
 
+                    config_data["my_wenxinworkshop"]["app_token"] = input_my_wenxinworkshop_app_token.value
+
                 if config.get("webui", "show_card", "llm", "gemini"):
                     config_data["gemini"]["api_key"] = input_gemini_api_key.value
                     config_data["gemini"]["model"] = select_gemini_model.value
@@ -2567,35 +2570,25 @@ def common_textarea_handle(content):
                         chatgpt_models = [
                             "gpt-3.5-turbo",
                             "gpt-3.5-turbo-instruct",
-                            "gpt-3.5-turbo-instruct-0914",
                             "gpt-3.5-turbo-0301",
                             "gpt-3.5-turbo-1106",
                             "gpt-3.5-turbo-0125",
                             "gpt-3.5-turbo-16k",
-                            "gpt-3.5-turbo-16k-0613",
                             "gpt-3.5-turbo-instruct",
-                            "gpt-3.5-turbo-instruct-0914",
                             "gpt-4",
                             "gpt-4-turbo-preview",
-                            "gpt-4-0613",
                             "gpt-4-32k",
-                            "gpt-4-32k-0613",
                             "gpt-4-1106-preview",
                             "gpt-4-0125-preview",
                             "text-embedding-3-large",
                             "text-embedding-3-small",
                             "text-davinci-003",
-                            "text-davinci-002",
-                            "text-curie-001",
-                            "text-babbage-001",
-                            "text-ada-001",
-                            "text-moderation-latest",
-                            "text-moderation-stable",
                             "rwkv",
                             "chatglm3-6b",
                             "moonshot-v1-8k",
                             "gemma:2b",
-                            "qwen"
+                            "qwen",
+                            "qwen:1.8b-chat"
                         ]
                         data_json = {}
                         for line in chatgpt_models:
@@ -2979,6 +2972,15 @@ def common_textarea_handle(content):
             if config.get("webui", "show_card", "llm", "my_wenxinworkshop"): 
                 with ui.card().style(card_css):
                     ui.label("千帆大模型")
+                    with ui.row():
+                        select_my_wenxinworkshop_type = ui.select(
+                            label='类型', 
+                            options={"千帆大模型": "千帆大模型", "AppBuilder": "AppBuilder"}, 
+                            value=config.get("my_wenxinworkshop", "type")
+                        ).style("width:150px")
+                        switch_my_wenxinworkshop_history_enable = ui.switch('上下文记忆', value=config.get("my_wenxinworkshop", "history_enable")).style(switch_internal_css)
+                        input_my_wenxinworkshop_history_max_len = ui.input(label='最大记忆长度', value=config.get("my_wenxinworkshop", "history_max_len"), placeholder='最长能记忆的问答字符串长度，超长会丢弃最早记忆的内容，请慎用！配置过大可能会有丢大米')
+
                     with ui.row():
                         input_my_wenxinworkshop_api_key = ui.input(label='api_key', value=config.get("my_wenxinworkshop", "api_key"), placeholder='千帆大模型平台，开通对应服务。应用接入-创建应用，填入api key')
                         input_my_wenxinworkshop_secret_key = ui.input(label='secret_key', value=config.get("my_wenxinworkshop", "secret_key"), placeholder='千帆大模型平台，开通对应服务。应用接入-创建应用，填入secret key')
@@ -3009,13 +3011,15 @@ def common_textarea_handle(content):
                             options=data_json, 
                             value=config.get("my_wenxinworkshop", "model")
                         ).style("width:150px")
-                        switch_my_wenxinworkshop_history_enable = ui.switch('上下文记忆', value=config.get("my_wenxinworkshop", "history_enable")).style(switch_internal_css)
-                        input_my_wenxinworkshop_history_max_len = ui.input(label='最大记忆长度', value=config.get("my_wenxinworkshop", "history_max_len"), placeholder='最长能记忆的问答字符串长度，超长会丢弃最早记忆的内容，请慎用！配置过大可能会有丢大米')
-                    with ui.row():
+
                         input_my_wenxinworkshop_temperature = ui.input(label='温度', value=config.get("my_wenxinworkshop", "temperature"), placeholder='(0, 1.0] 控制生成文本的随机性。较高的温度值会使生成的文本更随机和多样化，而较低的温度值会使生成的文本更加确定和一致。').style("width:200px;")
                         input_my_wenxinworkshop_top_p = ui.input(label='前p个选择', value=config.get("my_wenxinworkshop", "top_p"), placeholder='[0, 1.0] Nucleus采样。这个参数控制模型从累积概率大于一定阈值的令牌中进行采样。较高的值会产生更多的多样性，较低的值会产生更少但更确定的回答。').style("width:200px;")
                         input_my_wenxinworkshop_penalty_score = ui.input(label='惩罚得分', value=config.get("my_wenxinworkshop", "penalty_score"), placeholder='[1.0, 2.0] 在生成文本时对某些词语或模式施加的惩罚。这是一种调节生成内容的机制，用来减少或避免不希望出现的内容。').style("width:200px;")
+                    with ui.row():
+                        input_my_wenxinworkshop_app_token = ui.input(label='app_token', value=config.get("my_wenxinworkshop", "app_token"), placeholder='千帆AppBuilder平台，我的应用-应用配置-发布详情-我的Agent应用-API调用，填入app_token').style("width:200px;")
 
+
+
             # with ui.card().style(card_css):
             #     ui.label("千帆大模型（兼容问题暂不启用）")
             #     with ui.row():