Problems with NVIDIA driver
Hi!
Have you ever used a local generative language model?
What is the best way to use gemma-2-2b-it or any language model locally? Is there a C# tool to do that easily or do I better stick to Python?
I downloaded and installed Gemma-2-2b-it to my local computer (Asus VivoBook 17, 8GB ROM, intel code i3 processor) I followed this tutorial: https://youtu.be/7PbUaMdLGaA?si=uMpV_MSlM_8Ql7gs. I followed along the video tutorial, I copied testGemma.py in PyCharm and started the program, but it never gives an error, saying Found no NVIDIA driver on your system even though I have installed CUDA Toolkit 12.8 for WIN11, latest stable release. The program does not give me the answer. Any idea why? The pictures show the downloaded files and what I see in the IDE.
Aleksandar Haber PhD
YouTube
Install and Run Google's Gemma 2 (2B) small LLM in Python and Windo...
#gemma #google #machinelearning #ai #meta #llm #llama #llama31 #lamma3 #ai #machinelearning #largelanguagemodels
It takes a significant amount of time and energy to create these free video tutorials. You can support my efforts in this way:
- Buy me a Coffee: https://www.buymeacoffee.com/AleksandarHaber
- PayPal: https://www.paypal.me/Aleksandar...



6 Replies
Which laptop model exactly? Most of the Asus VivoBook 17 variants seem to only have Intel graphics. While it seems to be possible to get one with an NVIDIA GPU, I wouldn't have expected that to be paired with an i3
According to https://www.asus.com/uk/laptops/all-products?Spec=262000,264888, there's only 1 Vivobook 17 with 8GB RAM, and that comes with an i5, not an i3 (and has Intel graphics)
ASUS United Kingdom
All series|Laptops |ASUS United Kingdom
you can select Laptops which is most suitable for you according to the product features, specifications or price, and view detailed product information.Then also compare different products or add favorite item to your personal list in Laptops product list of ASUS "All series"
Alternatively, you can write your own wrapper over Ollama REST API (simplest) or switch to Semantic Kernel (more complex). All depends on your actual application to develop
ú

I really need to test some language models
So are these my only options?
- use CPU and wait for hours for a single prompt
- cloud GPU from Google Colab
- purchase and install a dedicated NVIDIA GPU that supports CUDA
- use optimized libraries like transformers and quantize models
https://www.asus.com/laptops/for-home/vivobook/vivobook-17-x712-11th-gen-intel/techspec/
The X712EA has Intel graphics, not an NVIDIA card
I doubt you'll be able to install a new GPU in that laptop, either
So using a cloud GPU is probably your best bet I reckon?