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 conversation API (#35)

Browse Source
This commit is contained in:
KevinHuSh
2024-01-18 19:28:37 +08:00
committed by GitHub
parent fad2ec7cf3
commit 4a858d33b6
13 changed files with 425 additions and 153 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
api/db/__init__.py
View File

@ -20,7 +20,7 @@ from strenum import StrEnum
class StatusEnum(Enum):
VALID = "1"
IN_VALID = "0"
INVALID = "0"
class UserTenantRole(StrEnum):

10
api/db/db_models.py
View File

@ -430,6 +430,7 @@ class LLM(DataBaseModel):
llm_name = CharField(max_length=128, null=False, help_text="LLM name", primary_key=True)
model_type = CharField(max_length=128, null=False, help_text="LLM, Text Embedding, Image2Text, ASR")
fid = CharField(max_length=128, null=False, help_text="LLM factory id")
max_tokens = IntegerField(default=0)
tags = CharField(max_length=255, null=False, help_text="LLM, Text Embedding, Image2Text, Chat, 32k...")
status = CharField(max_length=1, null=True, help_text="is it validate(0: wasted，1: validate)", default="1")
@ -467,8 +468,8 @@ class Knowledgebase(DataBaseModel):
doc_num = IntegerField(default=0)
token_num = IntegerField(default=0)
chunk_num = IntegerField(default=0)
similarity_threshold = FloatField(default=0.4)
vector_similarity_weight = FloatField(default=0.3)
#similarity_threshold = FloatField(default=0.4)
#vector_similarity_weight = FloatField(default=0.3)
parser_id = CharField(max_length=32, null=False, help_text="default parser ID")
status = CharField(max_length=1, null=True, help_text="is it validate(0: wasted，1: validate)", default="1")
@ -518,6 +519,11 @@ class Dialog(DataBaseModel):
prompt_type = CharField(max_length=16, null=False, default="simple", help_text="simple|advanced")
prompt_config = JSONField(null=False, default={"system": "", "prologue": "您好，我是您的助手小樱，长得可爱又善良，can I help you?",
"parameters": [], "empty_response": "Sorry! 知识库中未找到相关内容！"})
similarity_threshold = FloatField(default=0.4)
vector_similarity_weight = FloatField(default=0.3)
top_n = IntegerField(default=6)
kb_ids = JSONField(null=False, default=[])
status = CharField(max_length=1, null=True, help_text="is it validate(0: wasted，1: validate)", default="1")

12
api/db/init_data.py
View File

@ -62,61 +62,73 @@ def init_llm_factory():
"fid": factory_infos[0]["name"],
"llm_name": "gpt-3.5-turbo",
"tags": "LLM,CHAT,4K",
"max_tokens": 4096,
"model_type": LLMType.CHAT.value
},{
"fid": factory_infos[0]["name"],
"llm_name": "gpt-3.5-turbo-16k-0613",
"tags": "LLM,CHAT,16k",
"max_tokens": 16385,
"model_type": LLMType.CHAT.value
},{
"fid": factory_infos[0]["name"],
"llm_name": "text-embedding-ada-002",
"tags": "TEXT EMBEDDING,8K",
"max_tokens": 8191,
"model_type": LLMType.EMBEDDING.value
},{
"fid": factory_infos[0]["name"],
"llm_name": "whisper-1",
"tags": "SPEECH2TEXT",
"max_tokens": 25*1024*1024,
"model_type": LLMType.SPEECH2TEXT.value
},{
"fid": factory_infos[0]["name"],
"llm_name": "gpt-4",
"tags": "LLM,CHAT,8K",
"max_tokens": 8191,
"model_type": LLMType.CHAT.value
},{
"fid": factory_infos[0]["name"],
"llm_name": "gpt-4-32k",
"tags": "LLM,CHAT,32K",
"max_tokens": 32768,
"model_type": LLMType.CHAT.value
},{
"fid": factory_infos[0]["name"],
"llm_name": "gpt-4-vision-preview",
"tags": "LLM,CHAT,IMAGE2TEXT",
"max_tokens": 765,
"model_type": LLMType.IMAGE2TEXT.value
},{
"fid": factory_infos[1]["name"],
"llm_name": "qwen-turbo",
"tags": "LLM,CHAT,8K",
"max_tokens": 8191,
"model_type": LLMType.CHAT.value
},{
"fid": factory_infos[1]["name"],
"llm_name": "qwen-plus",
"tags": "LLM,CHAT,32K",
"max_tokens": 32768,
"model_type": LLMType.CHAT.value
},{
"fid": factory_infos[1]["name"],
"llm_name": "text-embedding-v2",
"tags": "TEXT EMBEDDING,2K",
"max_tokens": 2048,
"model_type": LLMType.EMBEDDING.value
},{
"fid": factory_infos[1]["name"],
"llm_name": "paraformer-realtime-8k-v1",
"tags": "SPEECH2TEXT",
"max_tokens": 25*1024*1024,
"model_type": LLMType.SPEECH2TEXT.value
},{
"fid": factory_infos[1]["name"],
"llm_name": "qwen_vl_chat_v1",
"tags": "LLM,CHAT,IMAGE2TEXT",
"max_tokens": 765,
"model_type": LLMType.IMAGE2TEXT.value
},
]

6
api/db/services/llm_service.py
View File

@ -34,7 +34,7 @@ class TenantLLMService(CommonService):
@classmethod
@DB.connection_context()
def get_api_key(cls, tenant_id, model_type):
def get_api_key(cls, tenant_id, model_type, model_name=""):
objs = cls.query(tenant_id=tenant_id, model_type=model_type)
if objs and len(objs)>0 and objs[0].llm_name:
return objs[0]
@ -42,7 +42,7 @@ class TenantLLMService(CommonService):
fields = [LLM.llm_name, cls.model.llm_factory, cls.model.api_key]
objs = cls.model.select(*fields).join(LLM, on=(LLM.fid == cls.model.llm_factory)).where(
(cls.model.tenant_id == tenant_id),
(cls.model.model_type == model_type),
((cls.model.model_type == model_type) | (cls.model.llm_name == model_name)),
(LLM.status == StatusEnum.VALID)
)
@ -60,7 +60,7 @@ class TenantLLMService(CommonService):
@classmethod
@DB.connection_context()
def model_instance(cls, tenant_id, llm_type):
model_config = cls.get_api_key(tenant_id, model_type=LLMType.EMBEDDING)
model_config = cls.get_api_key(tenant_id, model_type=LLMType.EMBEDDING.value)
if not model_config:
model_config = {"llm_factory": "local", "api_key": "", "llm_name": ""}
else: