The following playbook is no longer available: Try NVIDIA NIM APIs
The playbook is currently being reworked. We’ll publish and announce it a later time.
This was the main reason I bought the DGX Spark, to experiment with VLM / Nemotron and do fine tuning for inspection solutions. Until this playbook is rework are there any other examples or guides? I already have nemotron running, it’s the fine tuning I’m after/interested in.
I’ve shared a way of running nemotron VL NVFP4 on the spark. Follow the steps here: Running nvidia/Nemotron-Nano-VL-12B-V2-NVFP4-QAD on your spark - #4 by raphael.amorim
Thank you, great project, but I’m interested in fine tuning with my own data for specific use cases. Your project is great for using Nemotron, if you can do an enhancement on how to fine tune that would be greatly appreciated.
1 Like
Have you tried Nemotron Nano V2 VL — Megatron Bridge using Megatron Bridge — Megatron Bridge?
It’s using this recipe: Megatron-Bridge/examples/recipes/nemotron_vl/finetune_nemotron_nano_v2_vl.py at main · NVIDIA-NeMo/Megatron-Bridge · GitHub
1 Like
VLM playbook has been published: Live VLM WebUI | DGX Spark