Category: langchain

from langchain_experimental.text_splitter import SemanticChunker
from langchain_text_splitters import RecursiveCharacterTextSplitter


from langchain_community.document_loaders import DirectoryLoader
from langchain_community.embeddings import OllamaEmbeddings
from langchain_community.vectorstores import Chroma

# Load documents from a directory
loader = DirectoryLoader("./places_transcripts", glob="**/*.txt")

print("dir loaded loader")

documents = loader.load()

print(len(documents))

# # Create embeddingsclear
embeddings = OllamaEmbeddings(model="nomic-embed-text", show_progress=True)

# # Create Semantic Text Splitter
# text_splitter = SemanticChunker(embeddings, breakpoint_threshold_type="interquartile")

text_splitter = RecursiveCharacterTextSplitter(
    chunk_size=1500,
    chunk_overlap=300,
    add_start_index=True,
)

# # Split documents into chunks
texts = text_splitter.split_documents(documents)

# # Create vector store
vectorstore = Chroma.from_documents(
    documents=texts, 
    embedding= embeddings,
    persist_directory="./db-place")

print("vectorstore created")

from langchain_community.embeddings import OllamaEmbeddings
from langchain_community.vectorstores import Chroma
from langchain_community.chat_models import ChatOllama

from langchain.prompts import ChatPromptTemplate
from langchain.schema.runnable import RunnablePassthrough
from langchain.schema.output_parser import StrOutputParser


# # Create embeddingsclear
embeddings = OllamaEmbeddings(model="nomic-embed-text", show_progress=False)

db = Chroma(persist_directory="./db-place",
            embedding_function=embeddings)

# # Create retriever
retriever = db.as_retriever(
    search_type="similarity",
    search_kwargs= {"k": 5}
)

# # Create Ollama language model - Gemma 2
local_llm = 'phi3'

llm = ChatOllama(model=local_llm,
                 keep_alive="3h", 
                 max_tokens=512,  
                 temperature=0)

# Create prompt template
template = """<bos><start_of_turn>user\nAnswer the question based only on the following context and extract out a meaningful answer. \
Please write in full sentences with correct spelling and punctuation. if it makes sense use lists. \
If the context doen't contain the answer, just respond that you are unable to find an answer. \

CONTEXT: {context}

QUESTION: {question}

<end_of_turn>
<start_of_turn>model\n
ANSWER:"""
prompt = ChatPromptTemplate.from_template(template)

# Create the RAG chain using LCEL with prompt printing and streaming output
rag_chain = (
    {"context": retriever, "question": RunnablePassthrough()}
    | prompt
    | llm
)

# Function to ask questions
def ask_question(question):
    print("Answer:\n\n", end=" ", flush=True)
    for chunk in rag_chain.stream(question):
        print(chunk.content, end="", flush=True)
    print("\n")

# Example usage
if __name__ == "__main__":
    while True:
        user_question = input("Ask a question (or type 'quit' to exit): ")
        if user_question.lower() == 'quit':
            break
        answer = ask_question(user_question)
        # print("\nFull answer received.\n")