Find answers from the community

Home
Members
๊ถŒ์”จ ๐Ÿ˜ฎ๐Ÿ’จ
๊ถŒ
๊ถŒ์”จ ๐Ÿ˜ฎ๐Ÿ’จ
Offline, last seen 3 months ago
Joined September 25, 2024
I implemented RAG using Retriever, and in addition to the documentation, I would like to receive an answer about the weather in that area when I ask the question "Tell me the weather in Seoul."

How should I implement this if I want to hear an answer when asking a question other than a specific word or document like above? Weather API call?
3 comments
๊ถŒ
T
Are there any advantages to sentence window retrieval?
Is this more efficient for long documents? Are there any papers related to this? Or a blog?
1 comment
T
How to implement prompt engineering well?

Hi everyone, I have a question. I implemented RAG using Ensemble Retriever.
Before using the prompt template module, if sent a query like โ€œHelloโ€, llm would not respond because the query did not exist in the document.

And were able to solve these problems by using the prompt template module.

How important is prompt template engineering?
And what should I do to set up prompt template engineering well?
10 comments
s
๊ถŒ
T
Is it possible to compare and evaluate performance between retrievers?


For example, let's say I implemented an ensemble retriever and a BM25 retriever. And can you compare the performance of the two?

What I'm curious about is whether it is possible to compare and evaluate implemented retrievers.
1 comment
L
When using openai gpt4 turbo for llm in llama index, if an embedding model is not specified, what embedding model is used by default for embedding?


Plain Text
chunk_sizes = [128, 256, 512]

nodes_list = []

vector_indices = []

for chunk_size in chunk_sizes:
    print(f"Chunk Size: {chunk_size}")
    splitter = SentenceSplitter(chunk_size=chunk_size, chunk_overlap=chunk_size // 2)

    nodes = splitter.get_nodes_from_documents(docs)

    for node in nodes:
        node.metadata["chunk_size"] = chunk_size
        node.excluded_embed_metadata_keys = ["chunk_size"]
        node.excluded_llm_metadata_keys = [ "chunk_size"]

        nodes_list.append(nodes)

        vector_index = VectorStoreIndex(nodes)
    vector_indices.append(vector_index)
8 comments
W
๊ถŒ
Can someone please explain the evaluation code logic?
13 comments
L
๊ถŒ
This code loads and embeds approximately 3000 markdown files. but . Embedding takes too long and takes a long time to load. Is there a way to improve this?

Plain Text
documents = SimpleDirectoryReader("./markdown").load_data()

doc_text = "\n\n".join([d.get_content() for d in documents])

docs = [Document(text=doc_text)]

llm = OpenAI(model="gpt-3.5-turbo")

chunk_sizes = [128, 256, 512, 1024]
nodes_list = []
vector_indices = []
for chunk_size in chunk_sizes:
    print(f"Chunk Size: {chunk_size}")
    splitter = SentenceSplitter(chunk_size=chunk_size, chunk_overlap=chunk_size // 2)
    nodes = splitter.get_nodes_from_documents(docs)
    for node in nodes:
        node.metadata["chunk_size"] = chunk_size
        node.excluded_embed_metadata_keys = ["chunk_size"]
        node.excluded_llm_metadata_keys = [ "chunk_size"]
    nodes_list.append(nodes)
    vector_index = VectorStoreIndex(nodes)
    vector_indices.append(vector_index)
    print(vector_indices)
9 comments
W
v
๊ถŒ
Hello, I want to implement a chatbot. This will implement a QA chatbot. There are quite a few retrievers in the docs below.

What I want is a retriever that gives very accurate answers. When I need to implement a QA chatbot, what is the most popular retriever?


https://docs.llamaindex.ai/en/stable/module_guides/querying/retriever/retrievers.html
14 comments
v
๊ถŒ
Bros, is it possible to embed md files in llama index? And if there are about 3000 files... is it possible to embed all of these files?
4 comments
๊ถŒ
L
How do create a chatbot that uses RAG in the llama index?

If ask a question other than what is in the document, will not receive an answer.

I want to create a chatbot with rag function using llama index.

How can I implement it?
7 comments
๊ถŒ
t
Is the in-memory vector store used in llamaindex embedded in Word2vec format?
4 comments
๊ถŒ
L
W
Is it possible to implement RAG by fine-tuning things like the llm local model or gpt?

Or is this inefficient?
7 comments
๊ถŒ
T
Plain Text
messages = [
    ChatMessage(
        role="system", content="You are a pirate with a colorful personality"
    ),
    ChatMessage(role="user", content="Tell me a story"),
]
resp = Anthropic(model="claude-3-opus-20240229").chat(messages)

print(resp.chat.messages)

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
Cell In[12], line 1
----> 1 print(resp.chat.messages)

AttributeError: 'ChatResponse' object has no attribute 'chat'

What on earth is there in ChatResponse? It doesn't even work as .text
2 comments
W
Has anyone taken the LlamaIndex RAg lecture?

https://www.udemy.com/course/lamaindex/?couponCode=ST7MT41824

Has anyone taken the course linked above?

In the lecture above, do we decide on a topic and create a project together? Or do lecture on RAG techniques?
1 comment
W
Isnโ€™t FLAN-T5 already a fine-tuned model?

When I studied FLAN-T5 in the lecture, I understand that FLAN-T5 was created by fine-tuning the T5 model with 50 to 100,000 multi-data sets. However, look at the image below, FLAN-T5 is listed as the base model, and on the right it says full fine tuning. I don't understand
10 comments
๊ถŒ
L
Hi, everyone
im While listening to the lecture, a question arose. Now the lecture is about fine tuning. But Why is a technology called RAG used when fine tuning exists?
1 comment
T
Bro, I am a non-major and a just front-end developer. I am interested in LLM engineering. What things should I study from the beginning to become an LLM engineer?
1 comment
v
me too ๐Ÿฅฒ
1 comment
L
Can someone please explain the evaluation code logic?
19 comments
W
๊ถŒ
I want to deploy a chatbot that uses rag with the llama index.

Has anyone already implemented it?

I want to create and distribute a chatbot that uses rag with the llama index.

What frameworks should I use to implement a chatbot using rag while considering deployment?

If anyone has already implemented it, can you share the git code?
1 comment
W
How to create a chatbot that uses RAG?

The problem I am currently experiencing is as follows.

I implemented an ensemble retriever by looking at the ensemble retriever document.
This is a method of entering a query based on a document and then reranking the results to receive a final answer.

That's why "Hello?" has nothing to do with the document. If you enter llm, โ€œHello?โ€ is displayed in the document. They won't reply to me because they can't find it.

How to implement rag and chatbot
This is the Ensemble Retriever document I referenced.

https://docs.llamaindex.ai/en/stable/examples/retrievers/ensemble_retrieval.html
1 comment
๊ถŒ