支持软连接的知识库 (#2167)
* GLM3支持传入提示词 支持传入提示词的Agent ChatGLM3-6B * 软连接处理 1. 有效的软连接处理，参考了 #2125 2. 更新了依赖，现在fschat原生支持GLM3的全功能 * 更新了支持的模型列表和依赖 1. 推荐用户使用GLM3-6B取代2代GLM模型 2. 推荐用户使用性能更强的bge模型 3 推荐使用vllm和torch的最新稳定版本
This commit is contained in:
zR
GitHub
parent
dfcebf7bc3
commit
1b0cf67a57
|
|
@ -1,13 +1,12 @@
|
|||
import os
|
||||
|
||||
|
||||
# 可以指定一个绝对路径，统一存放所有的Embedding和LLM模型。
|
||||
# 每个模型可以是一个单独的目录，也可以是某个目录下的二级子目录。
|
||||
# 如果模型目录名称和 MODEL_PATH 中的 key 或 value 相同，程序会自动检测加载，无需修改 MODEL_PATH 中的路径。
|
||||
MODEL_ROOT_PATH = ""
|
||||
|
||||
# 选用的 Embedding 名称
|
||||
EMBEDDING_MODEL = "m3e-base" # bge-large-zh
|
||||
EMBEDDING_MODEL = "bge-large-zh"
|
||||
|
||||
# Embedding 模型运行设备。设为"auto"会自动检测，也可手动设定为"cuda","mps","cpu"其中之一。
|
||||
EMBEDDING_DEVICE = "auto"
|
||||
|
|
@ -18,7 +17,7 @@ EMBEDDING_MODEL_OUTPUT_PATH = "output"
|
|||
|
||||
# 要运行的 LLM 名称，可以包括本地模型和在线模型。
|
||||
# 第一个将作为 API 和 WEBUI 的默认模型
|
||||
LLM_MODELS = ["chatglm2-6b", "zhipu-api", "openai-api"]
|
||||
LLM_MODELS = ["chatglm3-6b", "zhipu-api", "openai-api"]
|
||||
|
||||
# AgentLM模型的名称 (可以不指定，指定之后就锁定进入Agent之后的Chain的模型，不指定就是LLM_MODELS[0])
|
||||
Agent_MODEL = None
|
||||
|
|
@ -153,6 +152,7 @@ MODEL_PATH = {
|
|||
"bge-large-zh-v1.5": "BAAI/bge-large-zh-v1.5",
|
||||
"piccolo-base-zh": "sensenova/piccolo-base-zh",
|
||||
"piccolo-large-zh": "sensenova/piccolo-large-zh",
|
||||
"nlp_gte_sentence-embedding_chinese-large": "damo/nlp_gte_sentence-embedding_chinese-large",
|
||||
"text-embedding-ada-002": "your OPENAI_API_KEY",
|
||||
},
|
||||
|
||||
|
|
@ -169,7 +169,7 @@ MODEL_PATH = {
|
|||
|
||||
"baichuan-7b": "baichuan-inc/Baichuan-7B",
|
||||
"baichuan-13b": "baichuan-inc/Baichuan-13B",
|
||||
'baichuan-13b-chat': 'baichuan-inc/Baichuan-13B-Chat',
|
||||
"baichuan-13b-chat": "baichuan-inc/Baichuan-13B-Chat",
|
||||
|
||||
"aquila-7b": "BAAI/Aquila-7B",
|
||||
"aquilachat-7b": "BAAI/AquilaChat-7B",
|
||||
|
|
@ -210,6 +210,12 @@ MODEL_PATH = {
|
|||
"Qwen-14B-Chat": "Qwen/Qwen-14B-Chat",
|
||||
"Qwen-14B-Chat-Int8": "Qwen/Qwen-14B-Chat-Int8", # 确保已经安装了auto-gptq optimum flash-attn
|
||||
"Qwen-14B-Chat-Int4": "Qwen/Qwen-14B-Chat-Int4", # 确保已经安装了auto-gptq optimum flash-attn
|
||||
|
||||
"agentlm-7b": "THUDM/agentlm-7b",
|
||||
"agentlm-13b": "THUDM/agentlm-13b",
|
||||
"agentlm-70b": "THUDM/agentlm-70b",
|
||||
|
||||
"Yi-34B-Chat": "https://huggingface.co/01-ai/Yi-34B-Chat", # 更多01-ai模型尚未进行测试。如果需要使用，请自行测试。
|
||||
},
|
||||
}
|
||||
|
||||
|
|
@ -225,7 +231,16 @@ VLLM_MODEL_DICT = {
|
|||
|
||||
"baichuan-7b": "baichuan-inc/Baichuan-7B",
|
||||
"baichuan-13b": "baichuan-inc/Baichuan-13B",
|
||||
'baichuan-13b-chat': 'baichuan-inc/Baichuan-13B-Chat',
|
||||
"baichuan-13b-chat": "baichuan-inc/Baichuan-13B-Chat",
|
||||
|
||||
"chatglm2-6b": "THUDM/chatglm2-6b",
|
||||
"chatglm2-6b-32k": "THUDM/chatglm2-6b-32k",
|
||||
"chatglm3-6b": "THUDM/chatglm3-6b",
|
||||
"chatglm3-6b-32k": "THUDM/chatglm3-6b-32k",
|
||||
|
||||
"BlueLM-7B-Chat": "vivo-ai/BlueLM-7B-Chat",
|
||||
"BlueLM-7B-Chat-32k": "vivo-ai/BlueLM-7B-Chat-32k",
|
||||
|
||||
# 注意：bloom系列的tokenizer与model是分离的，因此虽然vllm支持，但与fschat框架不兼容
|
||||
# "bloom":"bigscience/bloom",
|
||||
# "bloomz":"bigscience/bloomz",
|
||||
|
|
|
|||
|
|
@ -91,9 +91,12 @@ FSCHAT_MODEL_WORKERS = {
|
|||
|
||||
},
|
||||
# 可以如下示例方式更改默认配置
|
||||
# "baichuan-7b": { # 使用default中的IP和端口
|
||||
# "Qwen-7B-Chat": { # 使用default中的IP和端口
|
||||
# "device": "cpu",
|
||||
# },
|
||||
"chatglm3-6b": { # 使用default中的IP和端口
|
||||
"device": "cuda",
|
||||
},
|
||||
|
||||
#以下配置可以不用修改，在model_config中设置启动的模型
|
||||
"zhipu-api": {
|
||||
|
|
|
|||
|
|
@ -1,21 +1,21 @@
|
|||
# API requirements
|
||||
|
||||
langchain>=0.0.334
|
||||
langchain-experimental>=0.0.30
|
||||
fschat[model_worker]==0.2.32
|
||||
xformers>=0.0.22.post4
|
||||
openai>=0.28.1
|
||||
langchain-experimental>=0.0.42
|
||||
fschat[model_worker]>=0.2.33
|
||||
xformers>=0.0.22.post7
|
||||
openai~=0.28.1
|
||||
sentence_transformers
|
||||
transformers>=4.34
|
||||
torch>=2.0.1 # suggest version 2.1
|
||||
transformers>=4.35.2
|
||||
torch==2.1.0
|
||||
torchvision
|
||||
torchaudio
|
||||
fastapi>=0.104
|
||||
nltk>=3.8.1
|
||||
uvicorn~=0.23.1
|
||||
starlette~=0.27.0
|
||||
pydantic~=1.10.11
|
||||
unstructured[all-docs]>=0.10.12
|
||||
pydantic<2
|
||||
unstructured[all-docs]>=0.11.0
|
||||
python-magic-bin; sys_platform == 'win32'
|
||||
SQLAlchemy==2.0.19
|
||||
faiss-cpu
|
||||
|
|
@ -38,13 +38,14 @@ pandas~=2.0.3
|
|||
einops
|
||||
transformers_stream_generator==0.0.4
|
||||
|
||||
vllm>=0.2.0; sys_platform == "linux"
|
||||
vllm==0.2.2; sys_platform == "linux"
|
||||
|
||||
# online api libs
|
||||
# zhipuai
|
||||
# dashscope>=1.10.0 # qwen
|
||||
# qianfan
|
||||
# volcengine>=1.0.106 # fangzhou
|
||||
# online api libs dependencies
|
||||
|
||||
# zhipuai>=1.0.7
|
||||
# dashscope>=1.10.0
|
||||
# qianfan>=0.2.0
|
||||
# volcengine>=1.0.106
|
||||
|
||||
# uncomment libs if you want to use corresponding vector store
|
||||
# pymilvus==2.1.3 # requires milvus==2.1.3
|
||||
|
|
@ -53,10 +54,10 @@ vllm>=0.2.0; sys_platform == "linux"
|
|||
|
||||
# WebUI requirements
|
||||
|
||||
streamlit~=1.28.1
|
||||
streamlit~=1.28.2
|
||||
streamlit-option-menu>=0.3.6
|
||||
streamlit-antd-components>=0.1.11
|
||||
streamlit-antd-components>=0.2.3
|
||||
streamlit-chatbox>=1.1.11
|
||||
streamlit-aggrid>=0.3.4.post3
|
||||
httpx~=0.24.0
|
||||
httpx[brotli,http2,socks]~=0.24.1
|
||||
watchdog
|
||||
|
|
|
|||
|
|
@ -1,25 +1,25 @@
|
|||
# API requirements
|
||||
|
||||
langchain>=0.0.334
|
||||
langchain-experimental>=0.0.30
|
||||
fschat[model_worker]==0.2.32
|
||||
xformers>=0.0.22.post4
|
||||
openai>=0.28.1
|
||||
langchain-experimental>=0.0.42
|
||||
fschat[model_worker]>=0.2.33
|
||||
xformers>=0.0.22.post7
|
||||
openai~=0.28.1
|
||||
sentence_transformers
|
||||
transformers>=4.34
|
||||
torch>=2.0.1 # suggest version 2.1
|
||||
transformers>=4.35.2
|
||||
torch==2.1.0
|
||||
torchvision
|
||||
torchaudio
|
||||
fastapi>=0.104
|
||||
nltk>=3.8.1
|
||||
uvicorn~=0.23.1
|
||||
starlette~=0.27.0
|
||||
pydantic~=1.10.11
|
||||
unstructured[all-docs]>=0.10.12
|
||||
pydantic<2
|
||||
unstructured[all-docs]>=0.11.0
|
||||
python-magic-bin; sys_platform == 'win32'
|
||||
SQLAlchemy==2.0.19
|
||||
faiss-cpu
|
||||
accelerate
|
||||
accelerate>=0.24.1
|
||||
spacy
|
||||
PyMuPDF
|
||||
rapidocr_onnxruntime
|
||||
|
|
@ -36,7 +36,7 @@ websockets
|
|||
numpy~=1.24.4
|
||||
pandas~=2.0.3
|
||||
einops
|
||||
transformers_stream_generator==0.0.4
|
||||
transformers_stream_generator>=0.0.4
|
||||
|
||||
vllm>=0.2.0; sys_platform == "linux"
|
||||
|
||||
|
|
|
|||
|
|
@ -1,9 +1,10 @@
|
|||
# WebUI requirements
|
||||
|
||||
streamlit~=1.28.1
|
||||
streamlit~=1.28.2
|
||||
streamlit-option-menu>=0.3.6
|
||||
streamlit-antd-components>=0.1.11
|
||||
streamlit-antd-components>=0.2.3
|
||||
streamlit-chatbox>=1.1.11
|
||||
streamlit-aggrid>=0.3.4.post3
|
||||
httpx[brotli,http2,socks]>=0.25.0
|
||||
httpx[brotli,http2,socks]~=0.24.1
|
||||
watchdog
|
||||
|
||||
|
|
|
|||
|
|
@ -51,25 +51,35 @@ def list_kbs_from_folder():
|
|||
|
||||
|
||||
def list_files_from_folder(kb_name: str):
|
||||
def is_skiped_path(path: str): # 跳过 [temp, tmp, ., ~$] 开头的目录和文件
|
||||
tail = os.path.basename(path).lower()
|
||||
flag = False
|
||||
for x in ["temp", "tmp", ".", "~$"]:
|
||||
if tail.startswith(x):
|
||||
flag = True
|
||||
break
|
||||
return flag
|
||||
|
||||
doc_path = get_doc_path(kb_name)
|
||||
result = []
|
||||
for root, _, files in os.walk(doc_path):
|
||||
if is_skiped_path(root):
|
||||
continue
|
||||
for file in files:
|
||||
if is_skiped_path(file):
|
||||
continue
|
||||
path = Path(doc_path) / root / file
|
||||
result.append(path.resolve().relative_to(doc_path).as_posix())
|
||||
|
||||
def is_skiped_path(path: str):
|
||||
tail = os.path.basename(path).lower()
|
||||
for x in ["temp", "tmp", ".", "~$"]:
|
||||
if tail.startswith(x):
|
||||
return True
|
||||
return False
|
||||
|
||||
def process_entry(entry):
|
||||
if is_skiped_path(entry.path):
|
||||
return
|
||||
|
||||
if entry.is_symlink():
|
||||
target_path = os.path.realpath(entry.path)
|
||||
with os.scandir(target_path) as target_it:
|
||||
for target_entry in target_it:
|
||||
process_entry(target_entry)
|
||||
elif entry.is_file():
|
||||
result.append(entry.path)
|
||||
elif entry.is_dir():
|
||||
with os.scandir(entry.path) as it:
|
||||
for sub_entry in it:
|
||||
process_entry(sub_entry)
|
||||
|
||||
with os.scandir(doc_path) as it:
|
||||
for entry in it:
|
||||
process_entry(entry)
|
||||
|
||||
return result
|
||||
|
||||
|
|
|
|||
Loading…
Reference in New Issue