适配ai网关应用

2023-06-08 20:17:38 +08:00
parent 1ac5f934f2
commit 955367b0bc
4 changed files with 49 additions and 78 deletions
--- a/func_box.py
+++ b/func_box.py
@ -591,69 +591,5 @@ class JsonHandle:


 if __name__ == '__main__':
-    num = num_tokens_from_string([
-        """
-        You are jvs, 帮我做一篇关于chatgpt的分享,大纲为ChatGPT 的基本原理和应用、Prompt 的作用和优化、ChatGPT 和 Prompt 的案例研究、ChatGPT 和 Prompt 的挑战和限制、ChatGPT 和 Prompt 的未来发展
-Your decisions must always be made independently without seeking user assistance. Play to your strengths as an LLM and pursue simple strategies with no legal complications.
-GOALS:
-1. 在文后，将市面上的AI应用作比较，对每个部分进行适当调整和拓展
-Constraints:    # 确定目标
-1. ~4000 word limit for short term memory. Your short term memory is short, so immediately save important information to files.
-2. If you are unsure how you previously did something or want to recall past events, thinking about similar events will help you remember.
-3. No user assistance
-4. Exclusively use the commands listed below e.g. command_name
-Commands:   # 可执行的行动 
-1. analyze_code: Analyze Code, args: "code": "<full_code_string>"
-2. execute_python_file: Execute Python File, args: "filename": "<filename>"
-3. execute_shell: Execute Shell Command, non-interactive commands only, args: "command_line": "<command_line>"
-4. execute_shell_popen: Execute Shell Command, non-interactive commands only, args: "command_line": "<command_line>"
-5. append_to_file: Append to file, args: "filename": "<filename>", "text": "<text>"
-6. delete_file: Delete file, args: "filename": "<filename>"
-7. list_files: List Files in Directory, args: "directory": "<directory>"
-8. read_file: Read file, args: "filename": "<filename>"
-9. write_to_file: Write to file, args: "filename": "<filename>", "text": "<text>"
-10. google: Google Search, args: "query": "<query>"
-11. generate_image: Generate Image, args: "prompt": "<prompt>"
-12. improve_code: Get Improved Code, args: "suggestions": "<list_of_suggestions>", "code": "<full_code_string>"
-13. send_tweet: Send Tweet, args: "tweet_text": "<tweet_text>"
-14. browse_website: Browse Website, args: "url": "<url>", "question": "<what_you_want_to_find_on_website>"
-15. write_tests: Write Tests, args: "code": "<full_code_string>", "focus": "<list_of_focus_areas>"
-16. delete_agent: Delete GPT Agent, args: "key": "<key>"
-17. get_hyperlinks: Get text summary, args: "url": "<url>"
-18. get_text_summary: Get text summary, args: "url": "<url>", "question": "<question>"
-19. list_agents: List GPT Agents, args: () -> str
-20. message_agent: Message GPT Agent, args: "key": "<key>", "message": "<message>"
-21. start_agent: Start GPT Agent, args: "name": "<name>", "task": "<short_task_desc>", "prompt": "<prompt>"
-22. task_complete: Task Complete (Shutdown), args: "reason": "<reason>"
-Resources:         # 可用资源
-1. Internet access for searches and information gathering.
-2. Long Term memory management.
-3. GPT-3.5 powered Agents for delegation of simple tasks.
-4. File output.
-Performance Evaluation:               # 自我评估
-1. Continuously review and analyze your actions to ensure you are performing to the best of your abilities.
-2. Constructively self-criticize your big-picture behavior constantly.
-3. Reflect on past decisions and strategies to refine your approach.
-4. Every command has a cost, so be smart and efficient. Aim to complete tasks in the least number of steps.
-5. Write all code to a file.
-You should only respond in JSON format as described below  
-Response Format:             # 限制GPT的回答
-{
-    "thoughts": {
-        "text": "thought",
-        "reasoning": "reasoning",
-        "plan": "- short bulleted\n- list that conveys\n- long-term plan",
-        "criticism": "constructive self-criticism",
-        "speak": "thoughts summary to say to user"
-    },
-    "command": {
-        "name": "command name",
-        "args": {
-            "arg name": "value"
-        }
-    }
-} 
-Ensure the response can be parsed by Python json.loads
-        """
-    ])
+    num = num_tokens_from_string(['你好', '您好，请问有什么可以帮助您的吗？'])
    print(num)
--- a/request_llm/bridge_all.py
+++ b/request_llm/bridge_all.py
@ -48,10 +48,11 @@ class LazyloadTiktoken(object):
        return encoder.decode(*args, **kwargs)

 # Endpoint 重定向
-API_URL_REDIRECT, = get_conf("API_URL_REDIRECT")
+API_URL_REDIRECT, PROXY_API_URL = get_conf("API_URL_REDIRECT", 'PROXY_API_URL')
 openai_endpoint = "https://api.openai.com/v1/chat/completions"
 api2d_endpoint = "https://openai.api2d.net/v1/chat/completions"
 newbing_endpoint = "wss://sydney.bing.com/sydney/ChatHub"
+proxy_endpoint = PROXY_API_URL
 # 兼容旧版的配置
 try:
    API_URL, = get_conf("API_URL")
@ -66,6 +67,7 @@ if api2d_endpoint in API_URL_REDIRECT: api2d_endpoint = API_URL_REDIRECT[api2d_e
 if newbing_endpoint in API_URL_REDIRECT: newbing_endpoint = API_URL_REDIRECT[newbing_endpoint]


+
 # 获取tokenizer
 tokenizer_gpt35 = LazyloadTiktoken("gpt-3.5-turbo")
 tokenizer_gpt4 = LazyloadTiktoken("gpt-4")
@ -93,6 +95,24 @@ model_info = {
        "token_cnt": get_token_num_gpt4,
    },

+    "proxy-gpt-4": {
+        "fn_with_ui": chatgpt_ui,
+        "fn_without_ui": chatgpt_noui,
+        "endpoint": proxy_endpoint.replace('%v', 'gpt-4'),
+        "max_token": 8192,
+        "tokenizer": tokenizer_gpt4,
+        "token_cnt": get_token_num_gpt4,
+    },
+
+    "proxy-gpt-4-32k": {
+        "fn_with_ui": chatgpt_ui,
+        "fn_without_ui": chatgpt_noui,
+        "endpoint": proxy_endpoint.replace('%v', 'gpt-4-32k'),
+        "max_token": 32000,
+        "tokenizer": tokenizer_gpt4,
+        "token_cnt": get_token_num_gpt4,
+    },
+
    # api_2d
    "api2d-gpt-3.5-turbo": {
        "fn_with_ui": chatgpt_ui,
--- a/request_llm/bridge_chatgpt.py
+++ b/request_llm/bridge_chatgpt.py
@ -208,7 +208,7 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
                    error_msg = chunk_decoded
                    if "reduce the length" in error_msg:
                        if len(history) >= 2: history[-1] = ""; history[-2] = "" # 清除当前溢出的输入：history[-2] 是本次输入, history[-1] 是本次输出
-                        history = clip_history(inputs=inputs, history=history, tokenizer=model_info[llm_kwargs['llm_model']]['tokenizer'], 
+                        history = clip_history(inputs=inputs, history=history, tokenizer=model_info[llm_kwargs['llm_model']]['tokenizer'],
                                               max_token_limit=(model_info[llm_kwargs['llm_model']]['max_token'])) # history至少释放二分之一
                        chatbot[-1] = (chatbot[-1][0], "[Local Message] Reduce the length. 本次输入过长, 或历史数据过长. 历史缓存数据已部分释放, 您可以请再次尝试. (若再次失败则更可能是因为输入过长.)")
                        # history = []    # 清除历史
@ -238,13 +238,18 @@ def generate_payload(inputs, llm_kwargs, history, system_prompt, stream):
    """
    if not is_any_api_key(llm_kwargs['api_key']):
        raise AssertionError("你提供了错误的API_KEY。\n\n1. 临时解决方案：直接在输入区键入api_key，然后回车提交。\n\n2. 长效解决方案：在config.py中配置。")
-
-    api_key = select_api_key(llm_kwargs['api_key'], llm_kwargs['llm_model'])
-
-    headers = {
-        "Content-Type": "application/json",
-        "Authorization": f"Bearer {api_key}"
-    }
+    if llm_kwargs['llm_model'].startswith('proxy-'):
+        api_key = select_api_key(llm_kwargs['api_key'], llm_kwargs['llm_model'])
+        headers = {
+            "Content-Type": "application/json",
+            "api-key": f"{api_key}"
+        }
+    else:
+        api_key = select_api_key(llm_kwargs['api_key'], llm_kwargs['llm_model'])
+        headers = {
+            "Content-Type": "application/json",
+            "Authorization": f"Bearer {api_key}"
+        }

    conversation_cnt = len(history) // 2

--- a/toolbox.py
+++ b/toolbox.py
@ -54,10 +54,13 @@ def ArgsGeneralWrapper(f):
        """"""
        # 引入一个有cookie的chatbot
        start_time = time.time()
+        encrypt, private = get_conf('switch_model')[0]['key']
+        private_key, = get_conf('private_key')
        cookies.update({
            'top_p':top_p,
            'temperature':temperature,
        })
+
        llm_kwargs = {
            'api_key': cookies['api_key'],
            'llm_model': llm_model,
@ -70,8 +73,6 @@ def ArgsGeneralWrapper(f):
        plugin_kwargs = {
            "advanced_arg": plugin_advanced_arg
        }
-        encrypt, private = get_conf('switch_model')[0]['key']
-        private_key = get_conf('private_key')[0]
        if private in models:
            if chatbot == []:
                chatbot.append([None, f'隐私模式, 你的对话记录无法被他人检索 <p style="display:none;">\n{private_key}\n{ipaddr.client.host}\n</p>'])
@ -573,6 +574,12 @@ def is_api2d_key(key):
    else:
        return False

+def is_proxy_key(key):
+    if 'proxy' in key:
+        return True
+    else:
+        return False
+
 def is_any_api_key(key):
    if ',' in key:
        keys = key.split(',')
@ -580,7 +587,7 @@ def is_any_api_key(key):
            if is_any_api_key(k): return True
        return False
    else:
-        return is_openai_api_key(key) or is_api2d_key(key)
+        return is_openai_api_key(key) or is_api2d_key(key) or is_proxy_key(key)

 def what_keys(keys):
    avail_key_list = {'OpenAI Key':0, "API2D Key":0}
@ -609,6 +616,10 @@ def select_api_key(keys, llm_model):
        for k in key_list:
            if is_api2d_key(k): avail_key_list.append(k)

+    if llm_model.startswith('proxy'):
+        for k in key_list:
+            if is_proxy_key(k): avail_key_list.append(k.replace('proxy-', ''))
+
    if len(avail_key_list) == 0:
        raise RuntimeError(f"您提供的api-key不满足要求，不包含任何可用于{llm_model}的api-key。您可能选择了错误的模型或请求源。")

@ -905,4 +916,3 @@ def objload(file='objdump.tmp'):
        return
    with open(file, 'rb') as f:
        return pickle.load(f)
-