Thanks for the question. Can you please tell me about how you would use Lora and qwen3-7b model, and how large is your input size ?

I do not believe in large context windows, I know that working size is usually under 10K tokens. In regards to LoRA, I was just curious how it works with fine-tunes.

I do not believe in large context windows, I know that working size is usually under 10K tokens. In regards to LoRA, I was just curious how it works with fine-tunes.

To support the model, I have to add it manually. Technically, models fine-tuned using LoRA can be added

@asehmi Check out the gemma3-12B sample with image https://github.com/Mega4alik/ollm/blob/main/example_multimodality.py Thanks!

Thanks for the example.
I assume that because the tokens/sec rate is very low, that the primary use cases are long running background tasks that don't require chat-like user input. So, agent-like tasks.