最好的Prompt管理和使用依然是 Class 和 Function - 继续让LLM和编程语言融合

class GenerateAnswer(dspy.Signature):"""Answer questions with short factoid answers."""context = dspy.InputField(desc="may contain relevant facts")question = dspy.InputField()answer = dspy.OutputField(desc="often between 1 and 5 words")

# Option 1: Pass minimal signature to ChainOfThought modulegenerate_answer = dspy.ChainOfThought("context, question -> answer")
# Option 2: Or pass full notation signature to ChainOfThought modulegenerate_answer = dspy.ChainOfThought(GenerateAnswer)
# Call the module on a particular input.pred = generate_answer(context = "Which meant learning Lisp, since in those days Lisp was regarded as the language of AI.", question = "What programming language did the author learn in college?")

from sglang import function, system, user, assistant, gen, set_default_backend, RuntimeEndpoint
@functiondef multi_turn_question(s, question_1, question_2):s += system("You are a helpful assistant.")s += user(question_1)s += assistant(gen("answer_1", max_tokens=256))s += user(question_2)s += assistant(gen("answer_2", max_tokens=256))
set_default_backend(RuntimeEndpoint("http://localhost:30000"))
state = multi_turn_question.run(question_1="What is the capital of the United States?",question_2="List two local attractions.",)

import osos.environ["RAY_DEDUP_LOGS"] = "0" 
import rayfrom byzerllm.utils.retrieval import ByzerRetrievalfrom byzerllm.utils.client import ByzerLLM,LLMRequest,LLMResponse,LLMHistoryItem,InferBackendfrom byzerllm.utils.client import Templates
ray.init(address="auto",namespace="default",ignore_reinit_error=True)
llm = ByzerLLM()llm.setup_num_workers(2).setup_gpus_per_worker(0)
llm.deploy(pretrained_model_type="saas/sparkdesk", udf_name="sparkdesk_chat", infer_params={"saas.appid":"xxxx","saas.api_key":"xxxx","saas.api_secret":"xxxx","saas.gpt_url":"wss://spark-api.xf-yun.com/v3.5/chat" })

@llm.prompt()def generate_answer(context:str,question:str)->str:'''Answer the question based on only the following context:{context}Question: {question}Answer:'''pass

context='''Byzer产品栈从底层存储到大模型管理和serving再到应用开发框架：1. Byzer-Retrieval, 一个支持向量+搜索的混合检索数据库。2. Byzer-LLM, 可以衔接SaaS/开源模型，能部署，可以统一入口。3. Byzer-Agent ,一套分布式 Agent 框架，做为应用开发框架4. ByzerPerf, 一套性能吞吐评测框架5. ByzerEvaluation, 一套大模型效果评测框架 （未开源）6. Byzer-SQL， 一套全SQL方言，支持ETL，数据分析等工作，并且支持大模型的管理和调用（model as UDF）,方便数据预处理。'''print(generate_answer(context=context,question="Byzer SQL是什么？"))

Byzer SQL 是一个全SQL方言，支持ETL、数据分析等工作，并且支持大模型的管理和调用（model as UDF），方便数据预处理。

import rayfrom byzerllm.utils.client import ByzerLLMimport byzerllm
ray.init(address="auto",namespace="default",ignore_reinit_error=True)
class RAG():def __init__(self):self.llm = ByzerLLM()self.llm.setup_template(model="sparkdesk_chat",template="auto")self.llm.setup_default_model_name("sparkdesk_chat")@byzerllm.prompt(lambda self: self.llm)def generate_answer(self,context:str,question:str)->str:'''Answer the question based on only the following context:{context}Question: {question}Answer:'''pass

t = RAG()print(t.generate_answer(context=context,question="Byzer SQL是什么？"))

import rayfrom byzerllm.utils.client import ByzerLLMimport byzerllm
ray.init(address="auto",namespace="default",ignore_reinit_error=True)
class RAG():def __init__(self):self.llm = ByzerLLM()self.llm.setup_template(model="sparkdesk_chat",template="auto")self.llm.setup_default_model_name("sparkdesk_chat")@byzerllm.prompt()def generate_answer(self,context:str,question:str)->str:'''Answer the question based on only the following context:{context}Question: {question}Answer:'''pass
t = RAG()print(t.generate_answer(context=context,question="Byzer SQL是什么？"))

Answer the question based on only the following context:
Byzer产品栈从底层存储到大模型管理和serving再到应用开发框架：1. Byzer-Retrieval, 一个支持向量+搜索的混合检索数据库。2. Byzer-LLM, 可以衔接SaaS/开源模型，能部署，可以统一入口。3. Byzer-Agent ,一套分布式 Agent 框架，做为应用开发框架4. ByzerPerf, 一套性能吞吐评测框架5. ByzerEvaluation, 一套大模型效果评测框架 （未开源）6. Byzer-SQL， 一套全SQL方言，支持ETL，数据分析等工作，并且支持大模型的管理和调用（model as UDF）,方便数据预处理。
Question: Byzer SQL是什么？Answer:

import rayimport functoolsimport inspectimport byzerllmimport pydantic
ray.init(address="auto",namespace="default",ignore_reinit_error=True)
class ByzerProductDesc(pydantic.BaseModel):byzer_retrieval: strbyzer_llm: strbyzer_agent: strbyzer_perf: strbyzer_evaluation: strbyzer_sql: str
class RAG():def __init__(self):self.llm = ByzerLLM()self.llm.setup_template(model="sparkdesk_chat",template="auto")self.llm.setup_default_model_name("sparkdesk_chat")@byzerllm.prompt(lambda self: self.llm)def generate_answer(self,context:str,question:str)->ByzerProductDesc:'''Answer the question based on only the following context:{context}Question: {question}Answer:'''pass
t = RAG()
byzer_product = t.generate_answer(context=context,question="Byzer 产品列表")print(byzer_product.byzer_sql)## output: 一套全SQL方言，支持ETL，数据分析等工作，并且支持大模型的管理和调用（model as UDF），方便数据预处理

import rayimport functoolsimport inspectimport byzerllmimport pydanticfrom byzerllm.utils.client import ByzerLLM
ray.init(address="auto",namespace="default",ignore_reinit_error=True)
data = {'name': 'Jane Doe','task_count': 3,'tasks': [{'name': 'Submit report', 'due_date': '2024-03-10'},{'name': 'Finish project', 'due_date': '2024-03-15'},{'name': 'Reply to emails', 'due_date': '2024-03-08'}]}

class RAG():def __init__(self):self.llm = ByzerLLM()self.llm.setup_template(model="sparkdesk_chat",template="auto")self.llm.setup_default_model_name("sparkdesk_chat")@byzerllm.prompt(render="jinja2")def generate_answer(self,name,task_count,tasks)->str:'''Hello {{ name }},
This is a reminder that you have {{ task_count }} pending tasks:{% for task in tasks %}- Task: {{ task.name }} | Due: {{ task.due_date }}{% endfor %}
Best regards,Your Reminder System'''pass
t = RAG()
response = t.generate_answer(**data)print(response)