admin/dify
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

feat: advanced prompt backend (#1301)

Browse Source 
Co-authored-by: takatost <takatost@gmail.com>
This commit is contained in:
Garfield Dai
2023-10-12 23:13:10 +08:00
committed by GitHub
parent 2d1cb076c6
commit 42a5b3ec17
61 changed files with 767 additions and 581 deletions
Download Patch File Download Diff File Expand all files Collapse all files

108
api/core/generator/llm_generator.py
View File

@@ -10,9 +10,8 @@ from core.model_providers.models.entity.model_params import ModelKwargs
from core.prompt.output_parser.rule_config_generator import RuleConfigGeneratorOutputParser
from core.prompt.output_parser.suggested_questions_after_answer import SuggestedQuestionsAfterAnswerOutputParser
from core.prompt.prompt_template import JinjaPromptTemplate, OutLinePromptTemplate
from core.prompt.prompts import CONVERSATION_TITLE_PROMPT, CONVERSATION_SUMMARY_PROMPT, INTRODUCTION_GENERATE_PROMPT, \
GENERATOR_QA_PROMPT
from core.prompt.prompt_template import PromptTemplateParser
from core.prompt.prompts import CONVERSATION_TITLE_PROMPT, GENERATOR_QA_PROMPT
class LLMGenerator:
@@ -44,78 +43,19 @@ class LLMGenerator:
return answer.strip()
@classmethod
def generate_conversation_summary(cls, tenant_id: str, messages):
max_tokens = 200
model_instance = ModelFactory.get_text_generation_model(
tenant_id=tenant_id,
model_kwargs=ModelKwargs(
max_tokens=max_tokens
)
)
prompt = CONVERSATION_SUMMARY_PROMPT
prompt_with_empty_context = prompt.format(context='')
prompt_tokens = model_instance.get_num_tokens([PromptMessage(content=prompt_with_empty_context)])
max_context_token_length = model_instance.model_rules.max_tokens.max
max_context_token_length = max_context_token_length if max_context_token_length else 1500
rest_tokens = max_context_token_length - prompt_tokens - max_tokens - 1
context = ''
for message in messages:
if not message.answer:
continue
if len(message.query) > 2000:
query = message.query[:300] + "...[TRUNCATED]..." + message.query[-300:]
else:
query = message.query
if len(message.answer) > 2000:
answer = message.answer[:300] + "...[TRUNCATED]..." + message.answer[-300:]
else:
answer = message.answer
message_qa_text = "\n\nHuman:" + query + "\n\nAssistant:" + answer
if rest_tokens - model_instance.get_num_tokens([PromptMessage(content=context + message_qa_text)]) > 0:
context += message_qa_text
if not context:
return '[message too long, no summary]'
prompt = prompt.format(context=context)
prompts = [PromptMessage(content=prompt)]
response = model_instance.run(prompts)
answer = response.content
return answer.strip()
@classmethod
def generate_introduction(cls, tenant_id: str, pre_prompt: str):
prompt = INTRODUCTION_GENERATE_PROMPT
prompt = prompt.format(prompt=pre_prompt)
model_instance = ModelFactory.get_text_generation_model(
tenant_id=tenant_id
)
prompts = [PromptMessage(content=prompt)]
response = model_instance.run(prompts)
answer = response.content
return answer.strip()
@classmethod
def generate_suggested_questions_after_answer(cls, tenant_id: str, histories: str):
output_parser = SuggestedQuestionsAfterAnswerOutputParser()
format_instructions = output_parser.get_format_instructions()
prompt = JinjaPromptTemplate(
template="{{histories}}\n{{format_instructions}}\nquestions:\n",
input_variables=["histories"],
partial_variables={"format_instructions": format_instructions}
prompt_template = PromptTemplateParser(
template="{{histories}}\n{{format_instructions}}\nquestions:\n"
)
_input = prompt.format_prompt(histories=histories)
prompt = prompt_template.format({
"histories": histories,
"format_instructions": format_instructions
})
try:
model_instance = ModelFactory.get_text_generation_model(
@@ -128,10 +68,10 @@ class LLMGenerator:
except ProviderTokenNotInitError:
return []
prompts = [PromptMessage(content=_input.to_string())]
prompt_messages = [PromptMessage(content=prompt)]
try:
output = model_instance.run(prompts)
output = model_instance.run(prompt_messages)
questions = output_parser.parse(output.content)
except LLMError:
questions = []
@@ -145,19 +85,21 @@ class LLMGenerator:
def generate_rule_config(cls, tenant_id: str, audiences: str, hoping_to_solve: str) -> dict:
output_parser = RuleConfigGeneratorOutputParser()
prompt = OutLinePromptTemplate(
template=output_parser.get_format_instructions(),
input_variables=["audiences", "hoping_to_solve"],
partial_variables={
"variable": '{variable}',
"lanA": '{lanA}',
"lanB": '{lanB}',
"topic": '{topic}'
},
validate_template=False
prompt_template = PromptTemplateParser(
template=output_parser.get_format_instructions()
)
_input = prompt.format_prompt(audiences=audiences, hoping_to_solve=hoping_to_solve)
prompt = prompt_template.format(
inputs={
"audiences": audiences,
"hoping_to_solve": hoping_to_solve,
"variable": "{{variable}}",
"lanA": "{{lanA}}",
"lanB": "{{lanB}}",
"topic": "{{topic}}"
},
remove_template_variables=False
)
model_instance = ModelFactory.get_text_generation_model(
tenant_id=tenant_id,
@@ -167,10 +109,10 @@ class LLMGenerator:
)
)
prompts = [PromptMessage(content=_input.to_string())]
prompt_messages = [PromptMessage(content=prompt)]
try:
output = model_instance.run(prompts)
output = model_instance.run(prompt_messages)
rule_config = output_parser.parse(output.content)
except LLMError as e:
raise e