ivanwu/jarvis-models
1
0
Fork 0
mirror of https://github.com/BoardWare-Genius/jarvis-models.git synced 2025-12-13 16:53:24 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #11 from BoardWare-Genius/tom

Browse Source 
fix：blackbox chat
This commit is contained in:
ACBBZ
2024-05-29 17:30:38 +08:00
committed by GitHub
parent 6d80dd61e8 a049a60cb0
commit 381e0398f6
6 changed files with 232 additions and 78 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
requirements.txt
View File

@ -5,3 +5,8 @@ uvicorn==0.29.0
SpeechRecognition==3.10.3
PyYAML==6.0.1
injector==0.21.0
chromadb==0.5.0
langchain==0.1.17
langchain-community==0.0.36
sentence-transformers==2.7.0
openai

68
src/blackbox/chat.py
View File

@ -40,6 +40,9 @@ class Chat(Blackbox):
user_stop = settings.get("stop")
user_frequency_penalty = settings.get("frequency_penalty")
user_presence_penalty = settings.get("presence_penalty")
user_model_url = settings.get("model_url")
user_model_key = settings.get("model_key")
if user_context == None:
user_context = []
@ -72,16 +75,59 @@ class Chat(Blackbox):
if user_presence_penalty is None or user_presence_penalty == "":
user_presence_penalty = 0.8
if user_model_url is None or user_model_url.isspace() or user_model_url == "":
user_model_url = "http://120.196.116.194:48892/v1/chat/completions"
if user_model_key is None or user_model_key.isspace() or user_model_key == "":
user_model_key = "YOUR_API_KEY"
# 文心格式和openai的不一样，需要单独处理
if re.search(r"ernie", user_model_name):
# key = "24.22873ef3acf61fb343812681e4df251a.2592000.1719453781.282335-46723715" 没充钱，只有ernie-speed-128k能用
key = user_model_key
if re.search(r"ernie-speed-128k", user_model_name):
url = "https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie-speed-128k?access_token=" + key
elif re.search(r"ernie-3.5-8k", user_model_name):
url = "https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions?access_token=" + key
elif re.search(r"ernie-4.0-8k", user_model_name):
url = "https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions_pro?access_token=" + key
payload = json.dumps({
"system": prompt_template,
"messages": user_context + [
{
"role": "user",
"content": user_question
}
],
"temperature": user_temperature,
"top_p": user_top_p,
"stop": [str(user_stop)],
"max_output_tokens": user_max_tokens
})
headers = {
'Content-Type': 'application/json'
}
response = requests.request("POST", url, headers=headers, data=payload)
return response.json()["result"]
# gpt-4， gpt-3.5-turbo
if re.search(r"gpt", user_model_name):
elif re.search(r"gpt", user_model_name):
url = 'https://api.openai.com/v1/completions'
key = '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'
# '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'
key = user_model_key
# 自定义model
else:
url = 'http://120.196.116.194:48892/v1/chat/completions'
key = 'YOUR_API_KEY'
url = user_model_url
key = user_model_key
prompt_template = [
{"role": "system", "content": user_template},
]
@ -94,13 +140,13 @@ class Chat(Blackbox):
"content": user_question
}
],
"temperature": user_temperature,
"top_p": user_top_p,
"n": user_n,
"max_tokens": user_max_tokens,
"frequency_penalty": user_frequency_penalty,
"presence_penalty": user_presence_penalty,
"stop": user_stop
"temperature": str(user_temperature),
"top_p": str(user_top_p),
"n": str(user_n),
"max_tokens": str(user_max_tokens),
"frequency_penalty": str(user_frequency_penalty),
"presence_penalty": str(user_presence_penalty),
"stop": str(user_stop)
}
header = {

60
src/blackbox/chroma_chat.py
View File

@ -6,7 +6,10 @@ from .blackbox import Blackbox
from .chat import Chat
from .chroma_query import ChromaQuery
from ..log.logging_time import logging_time
import logging
logger = logging.getLogger
DEFAULT_COLLECTION_ID = "123"
from injector import singleton,inject
@ -24,19 +27,51 @@ class ChromaChat(Blackbox):
def valid(self, *args, **kwargs) -> bool:
data = args[0]
return isinstance(data, list)
@logging_time(logger=logger)
def processing(self, question: str, context: list, settings: dict) -> str:
def processing(self, question, context: list) -> str:
if context == None:
context = []
# chroma_chat settings
# {
# "chroma_embedding_model": "/model/Weight/BAAI/bge-large-zh-v1.5",
# "chroma_host": "10.6.82.192",
# "chroma_port": "8000",
# "chroma_collection_id": "g2e",
# "chroma_n_results": 3,
# "model_name": "Qwen1.5-14B-Chat",
# "context": [],
# "template": "",
# "temperature": 0,
# "top_p": 0.1,
# "n": 1,
# "max_tokens": 1024,
# "frequency_penalty": 0,
# "presence_penalty": 0,
# "stop": 100,
# "model_url": "http://120.196.116.194:48892/v1/chat/completions",
# "model_key": "YOUR_API_KEY"
# }
if settings is None:
settings = {}
# load or create collection
collection_id = DEFAULT_COLLECTION_ID
user_context = context
user_question = question
# query it
chroma_result = self.chroma_query(question, collection_id)
if user_context == None:
user_context = []
fast_question = "问题: "+ question + "。根据问题，总结以下内容和来源：" + chroma_result
response = self.chat(model_name="Qwen1.5-14B-Chat", prompt=fast_question, template='', context=context, temperature=0.8, top_p=0.8, n=1, max_tokens=1024, stop=100,frequency_penalty=0.5,presence_penalty=0.8)
if user_question is None:
return JSONResponse(content={"error": "question is required"}, status_code=status.HTTP_400_BAD_REQUEST)
# chroma answer
chroma_result = self.chroma_query(user_question, settings)
# chat prompt
fast_question = f"问题: {user_question}。- 根据知识库内的检索结果，以清晰简洁的表达方式回答问题。只从检索的内容中选取与问题相关信息。- 不要编造答案，如果答案不在经核实的资料中或无法从经核实的资料中得出，请回答“我无法回答您的问题。”检索内容：{chroma_result}"
# chat answer
response = self.chat(fast_question, user_context, settings)
return response
@ -49,10 +84,7 @@ class ChromaChat(Blackbox):
user_question = data.get("question")
user_context = data.get("context")
if user_question is None:
return JSONResponse(content={"error": "question is required"}, status_code=status.HTTP_400_BAD_REQUEST)
setting: dict = data.get("settings")
return JSONResponse(
content={"response": self.processing(user_question, user_context)},
status_code=status.HTTP_200_OK)
content={"response": self.processing(user_question, user_context, setting)}, status_code=status.HTTP_200_OK)

78
src/blackbox/chroma_query.py
View File

@ -6,8 +6,11 @@ from .blackbox import Blackbox
import chromadb
from chromadb.utils import embedding_functions
from ..utils import chroma_setting
import logging
from ..log.logging_time import logging_time
import re
logger = logging.getLogger
DEFAULT_COLLECTION_ID = "123"
from injector import singleton
@ -16,10 +19,11 @@ class ChromaQuery(Blackbox):
def __init__(self, *args, **kwargs) -> None:
# config = read_yaml(args[0])
# load embedding model
self.embedding_model = embedding_functions.SentenceTransformerEmbeddingFunction(model_name="/model/Weight/BAAI/bge-small-en-v1.5", device = "cuda")
# load chromadb
self.client = chromadb.HttpClient(host='10.6.82.192', port=8000)
# load chromadb and embedding model
self.embedding_model_1 = embedding_functions.SentenceTransformerEmbeddingFunction(model_name="/model/Weight/BAAI/bge-large-zh-v1.5", device = "cuda")
# self.embedding_model_2 = embedding_functions.SentenceTransformerEmbeddingFunction(model_name="/model/Weight/BAAI/bge-small-en-v1.5", device = "cuda")
self.client_1 = chromadb.HttpClient(host='10.6.82.192', port=8000)
# self.client_2 = chromadb.HttpClient(host='10.6.82.192', port=8000)
def __call__(self, *args, **kwargs):
return self.processing(*args, **kwargs)
@ -28,18 +32,62 @@ class ChromaQuery(Blackbox):
data = args[0]
return isinstance(data, list)
def processing(self, question: str, collection_id) -> str:
@logging_time(logger=logger)
def processing(self, question: str, settings: dict) -> str:
# load or create collection
collection = self.client.get_collection(collection_id, embedding_function=self.embedding_model)
if settings is None:
settings = {}
usr_question = question
# # chroma_query settings
chroma_embedding_model = settings.get("chroma_embedding_model")
chroma_host = settings.get("chroma_host")
chroma_port = settings.get("chroma_port")
chroma_collection_id = settings.get("chroma_collection_id")
chroma_n_results = settings.get("chroma_n_results")
if usr_question is None:
return JSONResponse(content={"error": "question is required"}, status_code=status.HTTP_400_BAD_REQUEST)
if chroma_embedding_model is None or chroma_embedding_model.isspace() or chroma_embedding_model == "":
chroma_embedding_model = "/model/Weight/BAAI/bge-large-zh-v1.5"
if chroma_host is None or chroma_host.isspace() or chroma_host == "":
chroma_host = "10.6.82.192"
if chroma_port is None or chroma_port.isspace() or chroma_port == "":
chroma_port = "8000"
if chroma_collection_id is None or chroma_collection_id.isspace() or chroma_collection_id == "":
chroma_collection_id = "g2e"
if chroma_n_results is None or chroma_n_results == "":
chroma_n_results = 3
# load client and embedding model from init
if re.search(r"10.6.82.192", chroma_host) and re.search(r"8000", chroma_port):
client = self.client_1
else:
client = chromadb.HttpClient(host=chroma_host, port=chroma_port)
if re.search(r"/model/Weight/BAAI/bge-large-zh-v1.5", chroma_embedding_model):
embedding_model = self.embedding_model_1
else:
embedding_model = embedding_functions.SentenceTransformerEmbeddingFunction(model_name=chroma_embedding_model, device = "cuda")
# load collection
collection = client.get_collection(chroma_collection_id, embedding_function=embedding_model)
# query it
results = collection.query(
query_texts=[question],
query_texts=[usr_question],
n_results=3,
)
response = str(results["documents"] + results["metadatas"])
# response = str(results["documents"] + results["metadatas"])
response = str(results["documents"])
return response
@ -50,14 +98,8 @@ class ChromaQuery(Blackbox):
return JSONResponse(content={"error": "json parse error"}, status_code=status.HTTP_400_BAD_REQUEST)
user_question = data.get("question")
user_collection_id = data.get("collection_id")
if user_question is None:
return JSONResponse(content={"error": "question is required"}, status_code=status.HTTP_400_BAD_REQUEST)
if user_collection_id is None:
user_collection_id = DEFAULT_COLLECTION_ID
setting = data.get("settings")
return JSONResponse(
content={"response": self.processing(user_question, user_collection_id)},
content={"response": self.processing(user_question, setting)},
status_code=status.HTTP_200_OK)

99
src/blackbox/chroma_upsert.py
View File

@ -17,7 +17,12 @@ import chromadb
import os
import tempfile
from ..utils import chroma_setting
import logging
from ..log.logging_time import logging_time
import re
logger = logging.getLogger
DEFAULT_COLLECTION_ID = "123"
from injector import singleton
@singleton
@ -26,9 +31,9 @@ class ChromaUpsert(Blackbox):
def __init__(self, *args, **kwargs) -> None:
# config = read_yaml(args[0])
# load embedding model
self.embedding_model = SentenceTransformerEmbeddings(model_name='/model/Weight/BAAI/bge-small-en-v1.5', model_kwargs={"device": "cuda"})
self.embedding_model_1 = SentenceTransformerEmbeddings(model_name="/model/Weight/BAAI/bge-large-zh-v1.5", model_kwargs={"device": "cuda"})
# load chroma db
self.client = chromadb.HttpClient(host='10.6.82.192', port=8000)
self.client_1 = chromadb.HttpClient(host='10.6.82.192', port=8000)
def __call__(self, *args, **kwargs):
return self.processing(*args, **kwargs)
@ -37,30 +42,59 @@ class ChromaUpsert(Blackbox):
data = args[0]
return isinstance(data, list)
def processing(self, collection_id, file, string, context, setting: chroma_setting) -> str:
@logging_time(logger=logger)
def processing(self, file, string, context: list, settings: dict) -> str:
# 用户的操作历史
if context is None:
context = []
context = [
{
"collection_id": "123",
"action": "query",
"content": "你吃饭了吗",
"answer": "吃了",
},
{
"collection_id": "123",
"action": "upsert",
"content": "file_name or string",
"answer": "collection 123 has 12472 documents. /tmp/Cheap and Quick：Efficient Vision-Language Instruction Tuning for Large Language Models.pdf ids is 0~111",
},
]
# context = [
# {
# "collection_id": "123",
# "action": "query",
# "content": "你吃饭了吗",
# "answer": "吃了",
# },
# {
# "collection_id": "123",
# "action": "upsert",
# "content": "file_name or string",
# "answer": "collection 123 has 12472 documents. /tmp/Cheap and Quick：Efficient Vision-Language Instruction Tuning for Large Language Models.pdf ids is 0~111",
# },
# ]
if collection_id is None and setting.ChromaSetting.collection_ids[0] != []:
collection_id = setting.ChromaSetting.collection_ids[0]
if settings is None:
settings = {}
# # chroma_query settings
chroma_embedding_model = settings.get("chroma_embedding_model")
chroma_host = settings.get("chroma_host")
chroma_port = settings.get("chroma_port")
chroma_collection_id = settings.get("chroma_collection_id")
if chroma_embedding_model is None or chroma_embedding_model.isspace() or chroma_embedding_model == "":
chroma_embedding_model = "/model/Weight/BAAI/bge-large-zh-v1.5"
if chroma_host is None or chroma_host.isspace() or chroma_host == "":
chroma_host = "10.6.82.192"
if chroma_port is None or chroma_port.isspace() or chroma_port == "":
chroma_port = "8000"
if chroma_collection_id is None or chroma_collection_id.isspace() or chroma_collection_id == "":
chroma_collection_id = "g2e"
# load client and embedding model from init
if re.search(r"10.6.82.192", chroma_host) and re.search(r"8000", chroma_port):
client = self.client_1
else:
collection_id = "123"
client = chromadb.HttpClient(host=chroma_host, port=chroma_port)
if re.search(r"/model/Weight/BAAI/bge-large-zh-v1.5", chroma_embedding_model):
embedding_model = self.embedding_model_1
else:
embedding_model = SentenceTransformerEmbeddings(model_name=chroma_embedding_model, device = "cuda")
if file is not None:
file_type = file.split(".")[-1]
@ -79,7 +113,6 @@ class ChromaUpsert(Blackbox):
loader = Docx2txtLoader(file)
elif file_type == "xlsx":
loader = UnstructuredExcelLoader(file)
documents = loader.load()
text_splitter = RecursiveCharacterTextSplitter(chunk_size=512, chunk_overlap=0)
@ -88,21 +121,21 @@ class ChromaUpsert(Blackbox):
ids = [str(file)+str(i) for i in range(len(docs))]
Chroma.from_documents(documents=docs, embedding=self.embedding_model, ids=ids, collection_name=collection_id, client=self.client)
Chroma.from_documents(documents=docs, embedding=embedding_model, ids=ids, collection_name=chroma_collection_id, client=client)
collection_number = self.client.get_collection(collection_id).count()
response_file = f"collection {collection_id} has {collection_number} documents. {file} ids is 0~{len(docs)-1}"
collection_number = client.get_collection(chroma_collection_id).count()
response_file = f"collection {chroma_collection_id} has {collection_number} documents. {file} ids is 0~{len(docs)-1}"
if string is not None:
# 生成一个新的id ids_string: 1
# ids = setting.ChromaSetting.string_ids[0] + 1
ids = "1"
Chroma.from_texts(texts=[string], embedding=self.embedding_model, ids=[ids], collection_name=collection_id, client=self.client)
Chroma.from_texts(texts=[string], embedding=embedding_model, ids=[ids], collection_name=chroma_collection_id, client=client)
collection_number = self.client.get_collection(collection_id).count()
response_string = f"collection {collection_id} has {collection_number} documents. {string} ids is {ids}"
collection_number = client.get_collection(chroma_collection_id).count()
response_string = f"collection {chroma_collection_id} has {collection_number} documents. {string} ids is {ids}"
if file is not None and string is not None:
@ -116,14 +149,10 @@ class ChromaUpsert(Blackbox):
async def fast_api_handler(self, request: Request) -> Response:
user_collection_id = (await request.form()).get("collection_id")
user_file = (await request.form()).get("file")
user_string = (await request.form()).get("string")
user_context = (await request.form()).get("context")
user_setting = (await request.form()).get("setting")
if user_collection_id is None and user_setting["collections"] == []:
return JSONResponse(content={"error": "The first creation requires a collection id"}, status_code=status.HTTP_400_BAD_REQUEST)
context = (await request.form()).get("context")
setting: dict = (await request.form()).get("settings")
if user_file is None and user_string is None:
return JSONResponse(content={"error": "file or string is required"}, status_code=status.HTTP_400_BAD_REQUEST)
@ -142,5 +171,5 @@ class ChromaUpsert(Blackbox):
return JSONResponse(
content={"response": self.processing(user_collection_id, safe_filename, user_string, user_context, user_setting)},
content={"response": self.processing(safe_filename, user_string, context, setting)},
status_code=status.HTTP_200_OK)

0
test.pdf
View File