Great job on this tutorial, thank you. I just have a question,I want to use llama-3-gguf as represenation model, but it takes so long to update the topics. I was wondering if it's normal or there's some thing I do wrong.
Thank you for the kind words! It's difficult to say without knowing which model size, quantization size, environment, GPU, the definition of "long", the code, etc. Could you provide that info?
I have used 'Meta-Llama-3-8B.Q4_K_M.gguf' model and I'm working in jupyter notebook.
Also for GPU, I use A100,80G. About the code, I don't know which part to share, I followed the steps you have provided in this tutorial and my code is almost identical. However, if there's any specific code I need to share, please let me know.
* What is your definition of "long to update the topics"?
* How many topics are there?
* Did you check whether the LLM was loaded onto the GPU?
> Meta-Llama-3-8B.Q4_K_M.gguf
It's not clear which model you are using purely based on this. What is the source of the model?
> About the code, I don't know which part to share, I followed the steps you have provided in this tutorial and my code is almost identical.
I would prefer it if you could share everything that involves creating the BERTopic model. Note that you can also simply open up an issue here: https://github.com/MaartenGr/BERTopic/issues
Great job on this tutorial, thank you. I just have a question,I want to use llama-3-gguf as represenation model, but it takes so long to update the topics. I was wondering if it's normal or there's some thing I do wrong.
Thank you for the kind words! It's difficult to say without knowing which model size, quantization size, environment, GPU, the definition of "long", the code, etc. Could you provide that info?
I have used 'Meta-Llama-3-8B.Q4_K_M.gguf' model and I'm working in jupyter notebook.
Also for GPU, I use A100,80G. About the code, I don't know which part to share, I followed the steps you have provided in this tutorial and my code is almost identical. However, if there's any specific code I need to share, please let me know.
Some questions:
* What is your definition of "long to update the topics"?
* How many topics are there?
* Did you check whether the LLM was loaded onto the GPU?
> Meta-Llama-3-8B.Q4_K_M.gguf
It's not clear which model you are using purely based on this. What is the source of the model?
> About the code, I don't know which part to share, I followed the steps you have provided in this tutorial and my code is almost identical.
I would prefer it if you could share everything that involves creating the BERTopic model. Note that you can also simply open up an issue here: https://github.com/MaartenGr/BERTopic/issues