andre/gpt_academic
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

适配ai网关应用

Browse Source
This commit is contained in:
w_xiaolizu
2023-06-08 20:17:38 +08:00
parent 1ac5f934f2
commit 955367b0bc
4 changed files with 49 additions and 78 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
request_llm/bridge_all.py
View File

@ -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,

21
request_llm/bridge_chatgpt.py
View File

@ -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