What's your go-to model? So many models…
Ha ha – not those models.
No matter how good your LLM server is, if the model you’re running isn't up to the task, then the results are going to be bad.
At the moment, I’m still testing out lots of different models, but I've found a nice balance with google/gemma-3-4b (3.03GB)
For more complex jobs maybe llama3.1 (8.55GB) or google/gemma-3-12b (8.07GB).
So, what’s working for you?
No comments:
Post a Comment