pixegami / rag-tutorial-v2

An Improved Langchain RAG Tutorial (v2) with local LLMs, database updates, and testing.

rag-tutorial-v2

rag-tutorial-v2's Issues

Using OpenAI Embeddings with `get_embedding_function.py`

HI Great video, I am trying to use OpenAI Embedding and keep running into a problem, can you help?

from openai import OpenAI
import os

client = OpenAI()

def get_embedding_function(text, model="text-embedding-3-small"):
   text = text.replace("\n", " ")
   return client.embeddings.create(input = [text], model=model).data[0].embedding

This is my updated code for the get_embedding_function.py script.

The embedding need input but i cannot see it being passed to it in the script in populate_database.py

Thanks in advance 😀 Jan

Add langchain_community to requirements

langchain_community is used for the Ollama model and the embedding function, but not included in the requirements.txt

LICENSE file?

Hi Jack,

First off, kudos on your excellent tutorial - it's been a real asset.
I'm eager to leverage it for a project I'm working on.
However, before I proceed, I noticed there isn't a LICENSE file attached to this repository.

Adding one would provide clarity on how your work can be used and built upon by others.
Would you mind considering this addition? It would be greatly appreciated.

Thanks again for your great work!

Best regards,
Chitoku

Not issue but need embedding for azure openai

def get_embedding_function():
embeddings = BedrockEmbeddings(
credentials_profile_name="default", region_name="us-east-1"
)
# embeddings = OllamaEmbeddings(model="nomic-embed-text")
return embeddings

Can we have similar function using Azure OPENAI
Wondering where is the text passed to this embedding? Sorry novice

Explicitly adding the id

In the video, he said something about explicitly adding the ids for the items, what does that mean and how do I do that?

Embedding failed

Followed the example & embedding is not working (give back empty sources)

If anyone has any clue would really appreciate it!

Problem: Check for Compatibility: Ensure that the version of langchain and langsmith you are using is compatible with the version of pydantic installed. Sometimes specific versions of these packages require specific versions of pydantic.

Haven't solved this yet but working on it.

C:\PycharmProjects\rag-tutorial-v2-main.venv\Scripts\python.exe C:/PyChar/plugins/python-ce/helpers/pycharm/_jb_pytest_runner.py --path C:\PycharmProjects\rag-tutorial-v2-main\test_rag.py
Testing started at 12:10 a.m. ...
Launching pytest with arguments C:\PycharmProjects\rag-tutorial-v2-main\test_rag.py --no-header --no-summary -q in C:\PycharmProjects\rag-tutorial-v2-main

============================= test session starts =============================
collecting ...
test_rag.py:None (test_rag.py)
test_rag.py:1: in
from query_data import query_rag
query_data.py:2: in
from langchain.vectorstores.chroma import Chroma
.venv\Lib\site-packages\langchain\vectorstores_init_.py:23: in
from langchain_core.vectorstores import VectorStore
.venv\Lib\site-packages\langchain_core\vectorstores.py:44: in
from langchain_core.retrievers import BaseRetriever
.venv\Lib\site-packages\langchain_core\retrievers.py:29: in
from langchain_core.documents import Document
.venv\Lib\site-packages\langchain_core\documents_init_.py:6: in
from langchain_core.documents.compressor import BaseDocumentCompressor
.venv\Lib\site-packages\langchain_core\documents\compressor.py:6: in
from langchain_core.callbacks import Callbacks
.venv\Lib\site-packages\langchain_core\callbacks_init_.py:22: in
from langchain_core.callbacks.manager import (
.venv\Lib\site-packages\langchain_core\callbacks\manager.py:29: in
from langsmith.run_helpers import get_run_tree_context
.venv\Lib\site-packages\langsmith\run_helpers.py:40: in
from langsmith import client as ls_client
.venv\Lib\site-packages\langsmith\client.py:52: in
from langsmith import env as ls_env
.venv\Lib\site-packages\langsmith\env_init_.py:3: in
from langsmith.env.runtime_env import (
.venv\Lib\site-packages\langsmith\env_runtime_env.py:10: in
from langsmith.utils import get_docker_compose_command
.venv\Lib\site-packages\langsmith\utils.py:31: in
from langsmith import schemas as ls_schemas
.venv\Lib\site-packages\langsmith\schemas.py:69: in
class Example(ExampleBase):
.venv\Lib\site-packages\pydantic\v1\main.py:286: in new
cls.try_update_forward_refs()
.venv\Lib\site-packages\pydantic\v1\main.py:807: in try_update_forward_refs
update_model_forward_refs(cls, cls.fields.values(), cls.config.json_encoders, localns, (NameError,))
.venv\Lib\site-packages\pydantic\v1\typing.py:554: in update_model_forward_refs
update_field_forward_refs(f, globalns=globalns, localns=localns)
.venv\Lib\site-packages\pydantic\v1\typing.py:520: in update_field_forward_refs
field.type = evaluate_forwardref(field.type_, globalns, localns or None)
.venv\Lib\site-packages\pydantic\v1\typing.py:66: in evaluate_forwardref
return cast(Any, type_)._evaluate(globalns, localns, set())
E TypeError: ForwardRef._evaluate() missing 1 required keyword-only argument: 'recursive_guard'
collected 0 items / 1 error

!!!!!!!!!!!!!!!!!!! Interrupted: 1 error during collection !!!!!!!!!!!!!!!!!!!!
============================== 1 error in 0.84s ===============================

Process finished with exit code 2

How do I invoke the AWS Bedrock model instead of the Ollama model in query_data.py?

I'd like to stick to using AWS for this project rather than Ollama.

The code that invokes the model in query_data.py is

model = Ollama(model="mistral")
response_text = model.invoke(prompt)

Any idea how I'd invoke the bedrock model in a similar manner? I've tried variations of the following but no luck yet...

    model_id = BedrockLLM(model_id="amazon.titan-embed-text-v1")
    response_text = model_id.invoke(prompt)