修改qianfan-api使用原始post访问，qianfan sdk无法访问
This commit is contained in:
liunux4odoo
parent
dea468ab0f
commit
b68f7fcdea
|
|
@ -1,68 +1,72 @@
|
||||||
import sys
|
import sys
|
||||||
from fastchat.conversation import Conversation
|
from fastchat.conversation import Conversation
|
||||||
from server.model_workers.base import *
|
from server.model_workers.base import *
|
||||||
|
from server.utils import get_httpx_client
|
||||||
|
from cachetools import cached, TTLCache
|
||||||
|
import json
|
||||||
from fastchat import conversation as conv
|
from fastchat import conversation as conv
|
||||||
import sys
|
import sys
|
||||||
from server.model_workers.base import ApiEmbeddingsParams
|
from server.model_workers.base import ApiEmbeddingsParams
|
||||||
from typing import List, Literal, Dict
|
from typing import List, Literal, Dict
|
||||||
|
|
||||||
|
|
||||||
# MODEL_VERSIONS = {
|
MODEL_VERSIONS = {
|
||||||
# "ernie-bot": "completions",
|
"ernie-bot-4": "completions_pro",
|
||||||
# "ernie-bot-turbo": "eb-instant",
|
"ernie-bot": "completions",
|
||||||
# "bloomz-7b": "bloomz_7b1",
|
"ernie-bot-turbo": "eb-instant",
|
||||||
# "qianfan-bloomz-7b-c": "qianfan_bloomz_7b_compressed",
|
"bloomz-7b": "bloomz_7b1",
|
||||||
# "llama2-7b-chat": "llama_2_7b",
|
"qianfan-bloomz-7b-c": "qianfan_bloomz_7b_compressed",
|
||||||
# "llama2-13b-chat": "llama_2_13b",
|
"llama2-7b-chat": "llama_2_7b",
|
||||||
# "llama2-70b-chat": "llama_2_70b",
|
"llama2-13b-chat": "llama_2_13b",
|
||||||
# "qianfan-llama2-ch-7b": "qianfan_chinese_llama_2_7b",
|
"llama2-70b-chat": "llama_2_70b",
|
||||||
# "chatglm2-6b-32k": "chatglm2_6b_32k",
|
"qianfan-llama2-ch-7b": "qianfan_chinese_llama_2_7b",
|
||||||
# "aquilachat-7b": "aquilachat_7b",
|
"chatglm2-6b-32k": "chatglm2_6b_32k",
|
||||||
# # "linly-llama2-ch-7b": "", # 暂未发布
|
"aquilachat-7b": "aquilachat_7b",
|
||||||
# # "linly-llama2-ch-13b": "", # 暂未发布
|
# "linly-llama2-ch-7b": "", # 暂未发布
|
||||||
# # "chatglm2-6b": "", # 暂未发布
|
# "linly-llama2-ch-13b": "", # 暂未发布
|
||||||
# # "chatglm2-6b-int4": "", # 暂未发布
|
# "chatglm2-6b": "", # 暂未发布
|
||||||
# # "falcon-7b": "", # 暂未发布
|
# "chatglm2-6b-int4": "", # 暂未发布
|
||||||
# # "falcon-180b-chat": "", # 暂未发布
|
# "falcon-7b": "", # 暂未发布
|
||||||
# # "falcon-40b": "", # 暂未发布
|
# "falcon-180b-chat": "", # 暂未发布
|
||||||
# # "rwkv4-world": "", # 暂未发布
|
# "falcon-40b": "", # 暂未发布
|
||||||
# # "rwkv5-world": "", # 暂未发布
|
# "rwkv4-world": "", # 暂未发布
|
||||||
# # "rwkv4-pile-14b": "", # 暂未发布
|
# "rwkv5-world": "", # 暂未发布
|
||||||
# # "rwkv4-raven-14b": "", # 暂未发布
|
# "rwkv4-pile-14b": "", # 暂未发布
|
||||||
# # "open-llama-7b": "", # 暂未发布
|
# "rwkv4-raven-14b": "", # 暂未发布
|
||||||
# # "dolly-12b": "", # 暂未发布
|
# "open-llama-7b": "", # 暂未发布
|
||||||
# # "mpt-7b-instruct": "", # 暂未发布
|
# "dolly-12b": "", # 暂未发布
|
||||||
# # "mpt-30b-instruct": "", # 暂未发布
|
# "mpt-7b-instruct": "", # 暂未发布
|
||||||
# # "OA-Pythia-12B-SFT-4": "", # 暂未发布
|
# "mpt-30b-instruct": "", # 暂未发布
|
||||||
# # "xverse-13b": "", # 暂未发布
|
# "OA-Pythia-12B-SFT-4": "", # 暂未发布
|
||||||
|
# "xverse-13b": "", # 暂未发布
|
||||||
|
|
||||||
# # # 以下为企业测试，需要单独申请
|
# # 以下为企业测试，需要单独申请
|
||||||
# # "flan-ul2": "",
|
# "flan-ul2": "",
|
||||||
# # "Cerebras-GPT-6.7B": ""
|
# "Cerebras-GPT-6.7B": ""
|
||||||
# # "Pythia-6.9B": ""
|
# "Pythia-6.9B": ""
|
||||||
# }
|
}
|
||||||
|
|
||||||
|
|
||||||
# @cached(TTLCache(1, 1800)) # 经过测试，缓存的token可以使用，目前每30分钟刷新一次
|
@cached(TTLCache(1, 1800)) # 经过测试，缓存的token可以使用，目前每30分钟刷新一次
|
||||||
# def get_baidu_access_token(api_key: str, secret_key: str) -> str:
|
def get_baidu_access_token(api_key: str, secret_key: str) -> str:
|
||||||
# """
|
"""
|
||||||
# 使用 AK，SK 生成鉴权签名（Access Token）
|
使用 AK，SK 生成鉴权签名（Access Token）
|
||||||
# :return: access_token，或是None(如果错误)
|
:return: access_token，或是None(如果错误)
|
||||||
# """
|
"""
|
||||||
# url = "https://aip.baidubce.com/oauth/2.0/token"
|
url = "https://aip.baidubce.com/oauth/2.0/token"
|
||||||
# params = {"grant_type": "client_credentials", "client_id": api_key, "client_secret": secret_key}
|
params = {"grant_type": "client_credentials", "client_id": api_key, "client_secret": secret_key}
|
||||||
# try:
|
try:
|
||||||
# with get_httpx_client() as client:
|
with get_httpx_client() as client:
|
||||||
# return client.get(url, params=params).json().get("access_token")
|
return client.get(url, params=params).json().get("access_token")
|
||||||
# except Exception as e:
|
except Exception as e:
|
||||||
# print(f"failed to get token from baidu: {e}")
|
print(f"failed to get token from baidu: {e}")
|
||||||
|
|
||||||
|
|
||||||
class QianFanWorker(ApiModelWorker):
|
class QianFanWorker(ApiModelWorker):
|
||||||
"""
|
"""
|
||||||
百度千帆
|
百度千帆
|
||||||
"""
|
"""
|
||||||
DEFAULT_EMBED_MODEL = "bge-large-zh"
|
DEFAULT_EMBED_MODEL = "embedding-v1"
|
||||||
|
|
||||||
def __init__(
|
def __init__(
|
||||||
self,
|
self,
|
||||||
|
|
@ -80,87 +84,98 @@ class QianFanWorker(ApiModelWorker):
|
||||||
|
|
||||||
def do_chat(self, params: ApiChatParams) -> Dict:
|
def do_chat(self, params: ApiChatParams) -> Dict:
|
||||||
params.load_config(self.model_names[0])
|
params.load_config(self.model_names[0])
|
||||||
import qianfan
|
# import qianfan
|
||||||
|
|
||||||
comp = qianfan.ChatCompletion(model=params.version,
|
# comp = qianfan.ChatCompletion(model=params.version,
|
||||||
endpoint=params.version_url,
|
# endpoint=params.version_url,
|
||||||
ak=params.api_key,
|
# ak=params.api_key,
|
||||||
sk=params.secret_key,)
|
# sk=params.secret_key,)
|
||||||
text = ""
|
|
||||||
for resp in comp.do(messages=params.messages,
|
|
||||||
temperature=params.temperature,
|
|
||||||
top_p=params.top_p,
|
|
||||||
stream=True):
|
|
||||||
if resp.code == 200:
|
|
||||||
if chunk := resp.body.get("result"):
|
|
||||||
text += chunk
|
|
||||||
yield {
|
|
||||||
"error_code": 0,
|
|
||||||
"text": text
|
|
||||||
}
|
|
||||||
else:
|
|
||||||
yield {
|
|
||||||
"error_code": resp.code,
|
|
||||||
"text": str(resp.body),
|
|
||||||
}
|
|
||||||
|
|
||||||
# BASE_URL = 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat'\
|
|
||||||
# '/{model_version}?access_token={access_token}'
|
|
||||||
|
|
||||||
# access_token = get_baidu_access_token(params.api_key, params.secret_key)
|
|
||||||
# if not access_token:
|
|
||||||
# yield {
|
|
||||||
# "error_code": 403,
|
|
||||||
# "text": f"failed to get access token. have you set the correct api_key and secret key?",
|
|
||||||
# }
|
|
||||||
|
|
||||||
# url = BASE_URL.format(
|
|
||||||
# model_version=params.version_url or MODEL_VERSIONS[params.version],
|
|
||||||
# access_token=access_token,
|
|
||||||
# )
|
|
||||||
# payload = {
|
|
||||||
# "messages": params.messages,
|
|
||||||
# "temperature": params.temperature,
|
|
||||||
# "stream": True
|
|
||||||
# }
|
|
||||||
# headers = {
|
|
||||||
# 'Content-Type': 'application/json',
|
|
||||||
# 'Accept': 'application/json',
|
|
||||||
# }
|
|
||||||
|
|
||||||
# text = ""
|
# text = ""
|
||||||
# with get_httpx_client() as client:
|
# for resp in comp.do(messages=params.messages,
|
||||||
# with client.stream("POST", url, headers=headers, json=payload) as response:
|
# temperature=params.temperature,
|
||||||
# for line in response.iter_lines():
|
# top_p=params.top_p,
|
||||||
# if not line.strip():
|
# stream=True):
|
||||||
# continue
|
# if resp.code == 200:
|
||||||
# if line.startswith("data: "):
|
# if chunk := resp.body.get("result"):
|
||||||
# line = line[6:]
|
# text += chunk
|
||||||
# resp = json.loads(line)
|
# yield {
|
||||||
|
# "error_code": 0,
|
||||||
|
# "text": text
|
||||||
|
# }
|
||||||
|
# else:
|
||||||
|
# yield {
|
||||||
|
# "error_code": resp.code,
|
||||||
|
# "text": str(resp.body),
|
||||||
|
# }
|
||||||
|
|
||||||
# if "result" in resp.keys():
|
BASE_URL = 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat'\
|
||||||
# text += resp["result"]
|
'/{model_version}?access_token={access_token}'
|
||||||
# yield {
|
|
||||||
# "error_code": 0,
|
access_token = get_baidu_access_token(params.api_key, params.secret_key)
|
||||||
# "text": text
|
if not access_token:
|
||||||
# }
|
yield {
|
||||||
# else:
|
"error_code": 403,
|
||||||
# yield {
|
"text": f"failed to get access token. have you set the correct api_key and secret key?",
|
||||||
# "error_code": resp["error_code"],
|
}
|
||||||
# "text": resp["error_msg"]
|
|
||||||
# }
|
url = BASE_URL.format(
|
||||||
|
model_version=params.version_url or MODEL_VERSIONS[params.version.lower()],
|
||||||
|
access_token=access_token,
|
||||||
|
)
|
||||||
|
payload = {
|
||||||
|
"messages": params.messages,
|
||||||
|
"temperature": params.temperature,
|
||||||
|
"stream": True
|
||||||
|
}
|
||||||
|
headers = {
|
||||||
|
'Content-Type': 'application/json',
|
||||||
|
'Accept': 'application/json',
|
||||||
|
}
|
||||||
|
|
||||||
|
text = ""
|
||||||
|
with get_httpx_client() as client:
|
||||||
|
with client.stream("POST", url, headers=headers, json=payload) as response:
|
||||||
|
for line in response.iter_lines():
|
||||||
|
if not line.strip():
|
||||||
|
continue
|
||||||
|
if line.startswith("data: "):
|
||||||
|
line = line[6:]
|
||||||
|
resp = json.loads(line)
|
||||||
|
|
||||||
|
if "result" in resp.keys():
|
||||||
|
text += resp["result"]
|
||||||
|
yield {
|
||||||
|
"error_code": 0,
|
||||||
|
"text": text
|
||||||
|
}
|
||||||
|
else:
|
||||||
|
yield {
|
||||||
|
"error_code": resp["error_code"],
|
||||||
|
"text": resp["error_msg"]
|
||||||
|
}
|
||||||
|
|
||||||
def do_embeddings(self, params: ApiEmbeddingsParams) -> Dict:
|
def do_embeddings(self, params: ApiEmbeddingsParams) -> Dict:
|
||||||
import qianfan
|
|
||||||
params.load_config(self.model_names[0])
|
params.load_config(self.model_names[0])
|
||||||
|
# import qianfan
|
||||||
|
|
||||||
embed = qianfan.Embedding(ak=params.api_key, sk=params.secret_key)
|
# embed = qianfan.Embedding(ak=params.api_key, sk=params.secret_key)
|
||||||
resp = embed.do(texts = params.texts, model=params.embed_model or self.DEFAULT_EMBED_MODEL)
|
# resp = embed.do(texts = params.texts, model=params.embed_model or self.DEFAULT_EMBED_MODEL)
|
||||||
if resp.code == 200:
|
# if resp.code == 200:
|
||||||
embeddings = [x.embedding for x in resp.body.get("data", [])]
|
# embeddings = [x.embedding for x in resp.body.get("data", [])]
|
||||||
return {"code": 200, "embeddings": embeddings}
|
# return {"code": 200, "embeddings": embeddings}
|
||||||
else:
|
# else:
|
||||||
return {"code": resp.code, "msg": str(resp.body)}
|
# return {"code": resp.code, "msg": str(resp.body)}
|
||||||
|
|
||||||
|
embed_model = params.embed_model or self.DEFAULT_EMBED_MODEL
|
||||||
|
access_token = get_baidu_access_token(params.api_key, params.secret_key)
|
||||||
|
url = f"https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/embeddings/{embed_model}?access_token={access_token}"
|
||||||
|
with get_httpx_client() as client:
|
||||||
|
resp = client.post(url, json={"input": params.texts}).json()
|
||||||
|
if "error_cdoe" not in resp:
|
||||||
|
embeddings = [x["embedding"] for x in resp.get("data", [])]
|
||||||
|
return {"code": 200, "embeddings": embeddings}
|
||||||
|
else:
|
||||||
|
return {"code": resp["error_code"], "msg": resp["error_msg"]}
|
||||||
|
|
||||||
|
|
||||||
# TODO: qianfan支持续写模型
|
# TODO: qianfan支持续写模型
|
||||||
|
|
|
||||||
|
|
@ -677,7 +677,7 @@ class ApiRequest:
|
||||||
return self._get_response_value(response, as_json=True, value_func=lambda r:r.get("data", []))
|
return self._get_response_value(response, as_json=True, value_func=lambda r:r.get("data", []))
|
||||||
|
|
||||||
|
|
||||||
def get_default_llm_model(self) -> (str, bool):
|
def get_default_llm_model(self) -> Tuple[str, bool]:
|
||||||
'''
|
'''
|
||||||
从服务器上获取当前运行的LLM模型，如果本机配置的LLM_MODEL属于本地模型且在其中，则优先返回
|
从服务器上获取当前运行的LLM模型，如果本机配置的LLM_MODEL属于本地模型且在其中，则优先返回
|
||||||
返回类型为（model_name, is_local_model）
|
返回类型为（model_name, is_local_model）
|
||||||
|
|
|
||||||
Loading…
Reference in New Issue