pip install dspy-ai[milvus]

from dspy.datasets import HotPotQA

# Load the dataset.
dataset = HotPotQA(train_seed=1, train_size=20, eval_seed=2023, dev_size=50, test_size=0)

# Tell DSPy that the 'question' field is the input. Any other fields are labels and/or metadata.
trainset = [x.with_inputs('question') for x in dataset.train]
devset = [x.with_inputs('question') for x in dataset.dev]

import requests
MILVUS_URI = "http://localhost:19530"
MILVUS_TOKEN = ""from pymilvus import MilvusClient, DataType, Collection
from dspy.retrieve.milvus_rm import openai_embedding_function

client = MilvusClient(
   uri=MILVUS_URI,
   token=MILVUS_TOKEN
)

if 'dspy_example' not in client.list_collections():
   client.create_collection(
      collection_name="dspy_example",
       overwrite= True,
       dimension=1536,
       primary_field_name="id",
       vector_field_name="embedding",
       id_type="int",
       metric_type="IP",
       max_length=65535,
       enable_dynamic=True
   )
text = requests.get('https://raw.githubusercontent.com/wxywb/dspy_dataset_sample/master/sample_data.txt').text

for idx, passage in enumerate(text.split('\n')):
   if len(passage) == 0:
       continue
   client.insert(collection_name="dspy_example", data = [{"id": idx , "embedding": openai_embedding_function(passage)[0], "text": passage}])

from dspy.retrieve.milvus_rm import MilvusRM
import os
import dspy
os.environ["OPENAI_API_KEY"] = "<YOUR_OPENAI_API_KEY>"
retriever_model = MilvusRM(collection_name="dspy_example",uri=MILVUS_URI,token=MILVUS_TOKEN,  # ignore this if no token is required for Milvus connection
   embedding_function = openai_embedding_function
   )
turbo = dspy.OpenAI(model='gpt-3.5-turbo')
dspy.settings.configure(lm=turbo)

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")

class RAG(dspy.Module):
 def __init__(self, rm):
 super().__init__()
 self.retrieve = rm

# This signature indicates the task imposed on the COT module.
 self.generate_answer = dspy.ChainOfThought(GenerateAnswer)

 def forward(self, question):
 # Use milvus_rm to retrieve context for the question.
 context = self.retrieve(question).passages
 # COT module takes "context, query" and output "answer".
 prediction = self.generate_answer(context=context, question=question)
 return dspy.Prediction(context=[item.long_text for item in context], answer=prediction.answer)

rag = RAG(retriever_model)
print(rag("who write At My Window").answer)

# The result:# 'Townes Van Zandt'

from dspy.evaluate.evaluate import Evaluate
from dspy.datasets import HotPotQA

evaluate_on_hotpotqa = Evaluate(devset=devset, num_threads=1, display_progress=False, display_table=5)

metric = dspy.evaluate.answer_exact_match
score = evaluate_on_hotpotqa(rag, metric=metric)
print('rag:', score)

# The result:# rag: 50.0

from dspy.teleprompt import BootstrapFewShot

# Validation logic: check that the predicted answer is correct.# Also check that the retrieved context does contain that answer.def validate_context_and_answer(example, pred, trace=None):
 answer_EM = dspy.evaluate.answer_exact_match(example, pred)
 answer_PM = dspy.evaluate.answer_passage_match(example, pred)
 return answer_EM and answer_PM

# Set up a basic teleprompter, which will compile our RAG program.
teleprompter = BootstrapFewShot(metric=validate_context_and_answer)

# Compile!
compiled_rag = teleprompter.compile(rag, trainset=trainset)

# Now compiled_rag is optimized and ready to answer your new question!

score = evaluate_on_hotpotqa(compiled_rag, metric=metric)
print(score)
print('compile_rag:', score)

# The result:# compile_rag: 52.0