Building and Evaluating Advanced RAG Applications - implementing it off platform

Has anyone been able to implement TruLens outside of this DeepLearning.AI platform? I tried to ask this question previously but got a somewhat helpful but somewhat ambiguous answer as to whether this is possible.

Just download the files (like etc.), pip install trulens and try it yourself. For me it works perfectly. Just be aware of potential costs with many evaluation questions if using gpt-3.5.

1 Like

David, Would you be able to confirm if you are succesfful to install trulens_eval as well and if so what version?

The reason I ask is that this fails for me if i do the off of platform with teh following code snippet and NameError

from trulens_eval import Tru
tru = Tru()

NameError: Fields must not use names with leading underscores; e.g., use ‘WithClassInfo__tru_class_info’ instead of ‘_WithClassInfo__tru_class_info’.

Thanks for any help.

The issue was seen in 0.18.2 of trulens_Eval and looks like it has regressed from 0.18.1. I could make it to work in 0.18.1

Yep, 0.18.1 in my case as well (with Python 3.11.4)

Thanks for the info. I was also able to get the local platform running using trulens_eval 0.18.1 and python 3.11.6.

I also saw this error (NameError: Fields must not use names with leading underscores; e.g., use ‘WithClassInfo__tru_class_info’ instead of ‘_WithClassInfo__tru_class_info’) last week. Not sure how I went around it. But now everything works fine with trulens-eval 0.18.2 (installed with pip) and python 3.11.6.

trulens-eval 0.18.3 works fine as well.

Hi All,
applying the code of the training on my local environment, running the following part:

from llama_index import SimpleDirectoryReader

documents = SimpleDirectoryReader(
input_files=[“/home/localpath/data/Knowledge Base v1.pdf”]

I receive the following error:
ModuleNotFoundError: No module named ‘langchain_core.tracers.langchain_v1’

I reinstalled langchain several times using conda and pip without solving the issue.

Conda virtualenv
Python 3.10.9

Do you have any suggestions?
Do you know the software packages required to run the code presented during the “Building and Evaluating Advanced RAG” course?

Thank you

I solve the problem using the following python version and packages:

Python 3.10.13
pip 23.3.1

Packages: (file)
python-dotenv 1.0.0
ipython 8.18.0
pypdf 3.16.1
openai 1.3.5
langchain 0.0.340
llama-index 0.9.8
trulens-eval 0.18.1
trulens 0.13.4
huggingface-hub 0.19.4
torch 2.1.1
torchvision 0.16.1

I hope this help who would like to implement a own solution on local systems.

Best Regards