Nvidia Nemotron 3 Family of Models

Re:Nvidia Nemotron 3 Family of Models by Tepix on Wednesday December 17, 2025 @05:55AM (#46300510)

Is it just me or is Nvidia trolling hard by calling a model with 30b parameters "nano"? With a bit of context, it doesn't even fit on a RTX 5090.

Other LLMs with the "nano" moniker are around 1b parameters or less.

[ Parent | on HN ]

Re:Nvidia Nemotron 3 Family of Models by patpatpat on Wednesday December 17, 2025 @08:42PM (#46308010)

FWIW It runs on my 9060xt(AMD) 16gb, without any tweaks just fine. It's very useable. I asked it to write a prime sieve in c#, started responding in .38 seconds, and wrote an implementation @ 20 tokens/sec

[ Parent | on HN ]

Re:Nvidia Nemotron 3 Family of Models by genpfault on Thursday December 18, 2025 @12:56AM (#46309334)

Getting ~150 tok/s on an empty context with a 24 GB 7900XTX via llama.cpp's Vukan backend.

[ Parent | on HN ]

Re:Nvidia Nemotron 3 Family of Models by Tepix on Saturday December 20, 2025 @04:29AM (#46334777)

Again, you're using some 3rd party quantisations, not the weights supplied by Nvidia (which don't fit in 24GB).

[ Parent | on HN ]

Re:Nvidia Nemotron 3 Family of Models by Tepix on Saturday December 20, 2025 @04:25AM (#46334752)

But you're using a 3rd party quant of unknown quality. Nvidia is only providing weights as BF16 and FP8.

[ Parent | on HN ]

Re:Nvidia Nemotron 3 Family of Models by barrystaes on Thursday December 18, 2025 @10:57AM (#46314309)

I wonder what performance remains on 12GB VRAM GPU when local ollama ties in the systems RAM to run this huge nano model.

https://github.com/jameschrisa/Ollama_Tuning_Guide/blob/main...

[ Parent | on HN ]