All You Need Is 4x 4090 GPUs to Train Your Own Model
sabareesh.com
All You Need Is 4x 4090 GPUs to Train Your Own Model
1–10 of 125 posts
Re: All You Need Is 4x 4090 GPUs to Train Your Own Model
#2Hey HN I am sharing my experience on how i pretrained my own LLM by building a ML rig at home
Re: All You Need Is 4x 4090 GPUs to Train Your Own Model
#3I would be much more intrested in a piece on what you can train with this kind of rig, rather than the rig itself
Re: All You Need Is 4x 4090 GPUs to Train Your Own Model
#4All you need is 4x 4090 GPUs to Train Your Own Model -- and $12000 to buy them
Re: All You Need Is 4x 4090 GPUs to Train Your Own Model
#5All you need is a 4x 4090 GPUs and a dedicated 30 amp circuit.
Re: All You Need Is 4x 4090 GPUs to Train Your Own Model
#6This article was written or rewritten via your model right?
The last paragraphs fell totally like AI.
Anyway I'd like a follow up on the curating, cleaning and training part which is far more interesting than how to select hardware which we've been doing for over 25 years.
Re: All You Need Is 4x 4090 GPUs to Train Your Own Model
#7It's probably better to hold out for the 5090 at this point, it's coming very soon as is expected to have 32GB of VRAM.
Re: All You Need Is 4x 4090 GPUs to Train Your Own Model
#8This article was written or rewritten via your model right? The last paragraphs fell totally like AI. Anyway I'd like a follow up on the curating, cleaning and training part which is far more interesting than how to select hardware which we've been doing for over 25 years.
Yes it is , thanks for the feedback. I will soon add it to github
Re: All You Need Is 4x 4090 GPUs to Train Your Own Model
#9It's probably better to hold out for the 5090 at this point, it's coming very soon as is expected to have 32GB of VRAM.
Yeah depends on the price, definitely 24GB is limiting
Re: All You Need Is 4x 4090 GPUs to Train Your Own Model
#10I would be much more intrested in a piece on what you can train with this kind of rig, rather than the rig itself
The bottleneck for most model training sizes is VRAM, and since each 4090 has 24 GB VRAM, that's 96 GB VRAM total. The article mentions that it can train LLMs from scratch up to 1 billion hyperparameters, which tracks.
Nowadays that's not a lot: a single H100 that you can now rent has 80 GB VRAM, and doesn't have the technical overhead of handling work across GPUs.