Spaces:
Sleeping
Sleeping
File size: 1,419 Bytes
317211f c0ab6fd 317211f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 |
from .model import InformationExtractedFromABillReceipt as PydanticModel
from langchain.chains import LLMChain
from langchain.chat_models import ChatOpenAI
from langchain.output_parsers import PydanticOutputParser, OutputFixingParser
from langchain.prompts import (
ChatPromptTemplate,
HumanMessagePromptTemplate,
SystemMessagePromptTemplate,
)
model = ChatOpenAI(
temperature=0,
n=1,
model_kwargs={
"stop": None,
"top_p": 1,
"frequency_penalty": 0,
"presence_penalty": 0,
},
)
# Build category chain
system_message_prompt = SystemMessagePromptTemplate.from_template(
"You are an information extraction engine that outputs details from OCR processed "
"documents like uids, total, tax, addresses, bank details, invoice details, "
"participant registration details. Fields with formats specified as date, time, or "
"datetime should be ISO 8601 compliant.\n"
"{format_instructions}"
)
human_message_prompt = HumanMessagePromptTemplate.from_template("{text}")
chat_prompt = ChatPromptTemplate.from_messages(
[system_message_prompt, human_message_prompt]
)
output_parser = PydanticOutputParser(pydantic_object=PydanticModel)
print(output_parser.get_format_instructions())
# exit()
fixing_parser = OutputFixingParser.from_llm(llm=model, parser=output_parser)
chain = LLMChain(llm=model, prompt=chat_prompt, output_parser=fixing_parser)
|