更新样式

app.py

@@ -18,8 +18,7 @@ with gr.Blocks(css=customCSS) as server:
         with gr.Column(scale=1):
             use_streaming_checkbox = gr.Checkbox(
                 label="实时传输回答", value=True, visible=enable_streaming_option)
-    chatbot = gr.Chatbot(elem_id="chat").style(
-        color_map=("#1D51EE", "#ffffff"))
+    chatbot = gr.Chatbot(elem_id="chat", color_map=("#1D51EE", "#ffffff"))
     history = gr.State([])
     token_count = gr.State([])
     promptTemplates = gr.State(load_template(

presets.py

@@ -1,7 +1,7 @@
 # -*- coding:utf-8 -*-
 title = """<h1 align="center">MyChatGPT</h1>"""
 description = """<div align=center>
-Powered by `gpt-3.5-turbo`
+Powered by `gpt-3.5-turbo` Model
 </div>
 """
 customCSS = """
@@ -39,5 +39,5 @@ max_token_streaming = 3500  # 流式对话时的最大 token 数
 timeout_streaming = 5  # 流式对话时的超时时间
 max_token_all = 3500  # 非流式对话时的最大 token 数
 timeout_all = 200  # 非流式对话时的超时时间
-enable_streaming_option = True  # 是否启用选择选择是否实时显示回答的勾选框
+enable_streaming_option = False  # 是否启用选择选择是否实时显示回答的勾选框
 HIDE_MY_KEY = True  # 如果你想在UI中隐藏你的 API 密钥，将此值设置为 True

utils.py

@@ -26,31 +26,34 @@ API_URL = "https://api.openai.com/v1/chat/completions"
 HISTORY_DIR = "history"
 TEMPLATES_DIR = "templates"
 
+
 def postprocess(
-        self, y: List[Tuple[str | None, str | None]]
-    ) -> List[Tuple[str | None, str | None]]:
-        """
-        Parameters:
-            y: List of tuples representing the message and response pairs. Each message and response should be a string, which may be in Markdown format.
-        Returns:
-            List of tuples representing the message and response. Each message and response will be a string of HTML.
-        """
-        if y is None:
-            return []
-        for i, (message, response) in enumerate(y):
-            y[i] = (
-                # None if message is None else markdown.markdown(message),
-                # None if response is None else markdown.markdown(response),
-                None if message is None else mdtex2html.convert((message)),
-                None if response is None else mdtex2html.convert(response),
-            )
-        return y
+    self, y: List[Tuple[str | None, str | None]]
+) -> List[Tuple[str | None, str | None]]:
+    """
+    Parameters:
+        y: List of tuples representing the message and response pairs. Each message and response should be a string, which may be in Markdown format.
+    Returns:
+        List of tuples representing the message and response. Each message and response will be a string of HTML.
+    """
+    if y is None:
+        return []
+    for i, (message, response) in enumerate(y):
+        y[i] = (
+            # None if message is None else markdown.markdown(message),
+            # None if response is None else markdown.markdown(response),
+            None if message is None else mdtex2html.convert((message)),
+            None if response is None else mdtex2html.convert(response),
+        )
+    return y
+
 
 def count_token(input_str):
     encoding = tiktoken.get_encoding("cl100k_base")
     length = len(encoding.encode(input_str))
     return length
 
+
 def parse_text(text):
     lines = text.split("\n")
     lines = [line for line in lines if line != ""]
@@ -82,21 +85,27 @@ def parse_text(text):
     text = "".join(lines)
     return text
 
+
 def construct_text(role, text):
     return {"role": role, "content": text}
 
+
 def construct_user(text):
     return construct_text("user", text)
 
+
 def construct_system(text):
     return construct_text("system", text)
 
+
 def construct_assistant(text):
     return construct_text("assistant", text)
 
+
 def construct_token_message(token, stream=False):
     return f"Token 计数: {token}"
 
+
 def get_response(openai_api_key, system_prompt, history, temperature, top_p, stream):
     headers = {
         "Content-Type": "application/json",
@@ -119,9 +128,11 @@ def get_response(openai_api_key, system_prompt, history, temperature, top_p, str
         timeout = timeout_streaming
     else:
         timeout = timeout_all
-    response = requests.post(API_URL, headers=headers, json=payload, stream=True, timeout=timeout)
+    response = requests.post(API_URL, headers=headers,
+                             json=payload, stream=True, timeout=timeout)
     return response
 
+
 def stream_predict(openai_api_key, system_prompt, history, inputs, chatbot, previous_token_count, top_p, temperature):
     def get_return_value():
         return chatbot, history, status_text, [*previous_token_count, token_counter]
@@ -140,7 +151,8 @@ def stream_predict(openai_api_key, system_prompt, history, inputs, chatbot, prev
         user_token_count = count_token(inputs)
     print(f"输入token计数: {user_token_count}")
     try:
-        response = get_response(openai_api_key, system_prompt, history, temperature, top_p, True)
+        response = get_response(
+            openai_api_key, system_prompt, history, temperature, top_p, True)
     except requests.exceptions.ConnectTimeout:
         status_text = standard_error_msg + error_retrieve_prompt
         yield get_return_value()
@@ -162,15 +174,19 @@ def stream_predict(openai_api_key, system_prompt, history, inputs, chatbot, prev
             # decode each line as response data is in bytes
             if chunklength > 6 and "delta" in chunk['choices'][0]:
                 finish_reason = chunk['choices'][0]['finish_reason']
-                status_text = construct_token_message(sum(previous_token_count)+token_counter+user_token_count, stream=True)
+                status_text = construct_token_message(
+                    sum(previous_token_count)+token_counter+user_token_count, stream=True)
                 if finish_reason == "stop":
                     print("生成完毕")
                     yield get_return_value()
                     break
                 try:
-                    partial_words = partial_words + chunk['choices'][0]["delta"]["content"]
+                    partial_words = partial_words + \
+                        chunk['choices'][0]["delta"]["content"]
                 except KeyError:
-                    status_text = standard_error_msg + "API回复中找不到内容��很可能是Token计数达到上限了。请重置对话。当前Token计数: " + str(sum(previous_token_count)+token_counter+user_token_count)
+                    status_text = standard_error_msg + "API回复中找不到内容。很可能是Token计数达到上限了。请重置对话。当前Token计数: " + \
+                        str(sum(previous_token_count) +
+                            token_counter+user_token_count)
                     yield get_return_value()
                     break
                 if token_counter == 0:
@@ -186,7 +202,8 @@ def predict_all(openai_api_key, system_prompt, history, inputs, chatbot, previou
     print("一次性回答模式")
     history.append(construct_user(inputs))
     try:
-        response = get_response(openai_api_key, system_prompt, history, temperature, top_p, False)
+        response = get_response(
+            openai_api_key, system_prompt, history, temperature, top_p, False)
     except requests.exceptions.ConnectTimeout:
         status_text = standard_error_msg + error_retrieve_prompt
         return chatbot, history, status_text, previous_token_count
@@ -201,26 +218,29 @@ def predict_all(openai_api_key, system_prompt, history, inputs, chatbot, previou
     return chatbot, history, status_text, previous_token_count
 
 
-def predict(openai_api_key, system_prompt, history, inputs, chatbot, token_count, top_p, temperature, stream=False, should_check_token_count = True):  # repetition_penalty, top_k
-    print("输入为：" +colorama.Fore.BLUE + f"{inputs}" + colorama.Style.RESET_ALL)
+def predict(openai_api_key, system_prompt, history, inputs, chatbot, token_count, top_p, temperature, stream=False, should_check_token_count=True):  # repetition_penalty, top_k
+    # print("输入为：" +colorama.Fore.BLUE + f"{inputs}" + colorama.Style.RESET_ALL)
     if stream:
         print("使用流式传输")
-        iter = stream_predict(openai_api_key, system_prompt, history, inputs, chatbot, token_count, top_p, temperature)
+        iter = stream_predict(openai_api_key, system_prompt, history,
+                              inputs, chatbot, token_count, top_p, temperature)
         for chatbot, history, status_text, token_count in iter:
             yield chatbot, history, status_text, token_count
     else:
         print("不使用流式传输")
-        chatbot, history, status_text, token_count = predict_all(openai_api_key, system_prompt, history, inputs, chatbot, token_count, top_p, temperature)
+        chatbot, history, status_text, token_count = predict_all(
+            openai_api_key, system_prompt, history, inputs, chatbot, token_count, top_p, temperature)
         yield chatbot, history, status_text, token_count
     print(f"传输完毕。当前token计数为{token_count}")
-    print("回答为：" +colorama.Fore.BLUE + f"{history[-1]['content']}" + colorama.Style.RESET_ALL)
+    # print("回答为：" +colorama.Fore.BLUE + f"{history[-1]['content']}" + colorama.Style.RESET_ALL)
     if stream:
         max_token = max_token_streaming
     else:
         max_token = max_token_all
     if sum(token_count) > max_token and should_check_token_count:
         print(f"精简token中{token_count}/{max_token}")
-        iter = reduce_token_size(openai_api_key, system_prompt, history, chatbot, token_count, top_p, temperature, stream=False, hidden=True)
+        iter = reduce_token_size(openai_api_key, system_prompt, history,
+                                 chatbot, token_count, top_p, temperature, stream=False, hidden=True)
         for chatbot, history, status_text, token_count in iter:
             status_text = f"Token 达到上限，已自动降低Token计数至 {status_text}"
             yield chatbot, history, status_text, token_count
@@ -234,7 +254,8 @@ def retry(openai_api_key, system_prompt, history, chatbot, token_count, top_p, t
     history.pop()
     inputs = history.pop()["content"]
     token_count.pop()
-    iter = predict(openai_api_key, system_prompt, history, inputs, chatbot, token_count, top_p, temperature, stream=stream)
+    iter = predict(openai_api_key, system_prompt, history, inputs,
+                   chatbot, token_count, top_p, temperature, stream=stream)
     print("重试完毕")
     for x in iter:
         yield x
@@ -242,7 +263,8 @@ def retry(openai_api_key, system_prompt, history, chatbot, token_count, top_p, t
 
 def reduce_token_size(openai_api_key, system_prompt, history, chatbot, token_count, top_p, temperature, stream=False, hidden=False):
     print("开始减少token数量……")
-    iter = predict(openai_api_key, system_prompt, history, summarize_prompt, chatbot, token_count, top_p, temperature, stream=stream, should_check_token_count=False)
+    iter = predict(openai_api_key, system_prompt, history, summarize_prompt, chatbot,
+                   token_count, top_p, temperature, stream=stream, should_check_token_count=False)
     for chatbot, history, status_text, previous_token_count in iter:
         history = history[-2:]
         token_count = previous_token_count[-1:]
@@ -309,9 +331,11 @@ def load_chat_history(filename, system, history, chatbot):
         print("没有找到对话历史文件，不执行任何操作")
         return filename, system, history, chatbot
 
+
 def sorted_by_pinyin(list):
     return sorted(list, key=lambda char: lazy_pinyin(char)[0][0])
 
+
 def get_file_names(dir, plain=False, filetypes=[".json"]):
     print(f"获取文件名列表，目录为{dir}，文件类型为{filetypes}，是否为纯文本列表{plain}")
     files = []
@@ -328,10 +352,12 @@ def get_file_names(dir, plain=False, filetypes=[".json"]):
     else:
         return gr.Dropdown.update(choices=files)
 
+
 def get_history_names(plain=False):
     print("获取历史记录文件名列表")
     return get_file_names(HISTORY_DIR, plain)
 
+
 def load_template(filename, mode=0):
     print(f"加载模板文件{filename}，模式为{mode}（0为返回字典和下拉菜单，1为返回下拉菜单，2为返回字典）")
     lines = []
@@ -348,15 +374,17 @@ def load_template(filename, mode=0):
     if mode == 1:
         return sorted_by_pinyin([row[0] for row in lines])
     elif mode == 2:
-        return {row[0]:row[1] for row in lines}
+        return {row[0]: row[1] for row in lines}
     else:
         choices = sorted_by_pinyin([row[0] for row in lines])
-        return {row[0]:row[1] for row in lines}, gr.Dropdown.update(choices=choices, value=choices[0])
+        return {row[0]: row[1] for row in lines}, gr.Dropdown.update(choices=choices, value=choices[0])
+
 
 def get_template_names(plain=False):
     print("获取模板文件名列表")
     return get_file_names(TEMPLATES_DIR, plain, filetypes=[".csv", "json"])
 
+
 def get_template_content(templates, selection, original_system_prompt):
     print(f"应用模板中，选择为{selection}，原始系统提示为{original_system_prompt}")
     try:
@@ -364,9 +392,11 @@ def get_template_content(templates, selection, original_system_prompt):
     except:
         return original_system_prompt
 
+
 def reset_state():
     print("重置状态")
     return [], [], [], construct_token_message(0)
 
+
 def reset_textbox():
     return gr.update(value='')