pengjunjie/ragflow
1
0
Fork 1
mirror of https://github.com/infiniflow/ragflow.git synced 2025-12-08 20:42:30 +08:00
Code Issues Packages Projects Releases Wiki Activity

Add 'One' chunk method (#137)

Browse Source
This commit is contained in:
KevinHuSh
2024-03-20 18:57:22 +08:00
committed by GitHub
parent fce14ee187
commit 5875c8ba08
11 changed files with 143 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
api/db/__init__.py
View File

@ -79,3 +79,4 @@ class ParserType(StrEnum):
TABLE = "table"
NAIVE = "naive"
PICTURE = "picture"
ONE = "one"

12
api/db/init_data.py
View File

@ -79,12 +79,12 @@ factory_infos = [{
"tags": "LLM,TEXT EMBEDDING,SPEECH2TEXT,MODERATION",
"status": "1",
},{
"name": "通义千问",
"name": "Tongyi-Qianwen",
"logo": "",
"tags": "LLM,TEXT EMBEDDING,SPEECH2TEXT,MODERATION",
"status": "1",
},{
"name": "智谱AI",
"name": "ZHIPU-AI",
"logo": "",
"tags": "LLM,TEXT EMBEDDING,SPEECH2TEXT,MODERATION",
"status": "1",
@ -270,6 +270,14 @@ def init_llm_factory():
except Exception as e:
pass
"""
drop table llm;
drop table factories;
update tenant_llm set llm_factory='Tongyi-Qianwen' where llm_factory='通义千问';
update tenant_llm set llm_factory='ZHIPU-AI' where llm_factory='智谱AI';
update tenant set parser_ids='naive:General,one:One,qa:Q&A,resume:Resume,table:Table,laws:Laws,manual:Manual,book:Book,paper:Paper,presentation:Presentation,picture:Picture';
"""
def init_web_data():
start_time = time.time()

12
api/settings.py
View File

@ -52,7 +52,7 @@ REQUEST_MAX_WAIT_SEC = 300
USE_REGISTRY = get_base_config("use_registry")
default_llm = {
"通义千问": {
"Tongyi-Qianwen": {
"chat_model": "qwen-plus",
"embedding_model": "text-embedding-v2",
"image2text_model": "qwen-vl-max",
@ -64,7 +64,7 @@ default_llm = {
"image2text_model": "gpt-4-vision-preview",
"asr_model": "whisper-1",
},
"智谱AI": {
"ZHIPU-AI": {
"chat_model": "glm-3-turbo",
"embedding_model": "embedding-2",
"image2text_model": "glm-4v",
@ -84,17 +84,17 @@ default_llm = {
}
}
LLM = get_base_config("user_default_llm", {})
LLM_FACTORY = LLM.get("factory", "通义千问")
LLM_FACTORY = LLM.get("factory", "Tongyi-Qianwen")
if LLM_FACTORY not in default_llm:
print("\33[91m【ERROR】\33[0m:", f"LLM factory {LLM_FACTORY} has not supported yet, switch to '通义千问/QWen' automatically, and please check the API_KEY in service_conf.yaml.")
LLM_FACTORY = "通义千问"
print("\33[91m【ERROR】\33[0m:", f"LLM factory {LLM_FACTORY} has not supported yet, switch to 'Tongyi-Qianwen/QWen' automatically, and please check the API_KEY in service_conf.yaml.")
LLM_FACTORY = "Tongyi-Qianwen"
CHAT_MDL = default_llm[LLM_FACTORY]["chat_model"]
EMBEDDING_MDL = default_llm[LLM_FACTORY]["embedding_model"]
ASR_MDL = default_llm[LLM_FACTORY]["asr_model"]
IMAGE2TEXT_MDL = default_llm[LLM_FACTORY]["image2text_model"]
API_KEY = LLM.get("api_key", "")
PARSERS = LLM.get("parsers", "naive:General,qa:Q&A,resume:Resume,table:Table,laws:Laws,manual:Manual,book:Book,paper:Paper,presentation:Presentation,picture:Picture")
PARSERS = LLM.get("parsers", "naive:General,one:One,qa:Q&A,resume:Resume,table:Table,laws:Laws,manual:Manual,book:Book,paper:Paper,presentation:Presentation,picture:Picture")
# distribution
DEPENDENT_DISTRIBUTION = get_base_config("dependent_distribution", False)