How much VRAM would this require, if I would want to run this locally?<p>I bought a 12GB Nvidia card a year ago. In general I'm having a hard time to find the actual required hardware specs for any self hosted AI model. Any tips/suggestions/recommended resources for that?
12GB will be sufficient to run a quantized version, provided you're not running anything else memory-hungry on the GPU.<p>You're not finding hardware specs because there are a lot of variables at play - the degree to which the weights are quantized, how much space you want to set aside for the KV cache, extra memory needed for multimodal features, etc.<p>My rule of thumb is 1 byte per parameter to be comfortable (running a quantization with somewhere between 4.5 and 6 bits per parameter and leaving some room for the cache and extras), so 7 GB for 7 billion parameters. If you need a really large context you'll need more; if you want to push it you can get away with a little less.
One quick way to estimate a lower bound is to take the number of parameters and multiply it with the bits per parameter. So a model with 7 billion parameters running with float8 types would be ~7 GB to load at a <i>minimum</i>. The attention mechanism would require more on top of that, and depends on the size of the context window.<p>You'll also need to load inputs (images in this case) onto the GPU memory, and that depends on the image resolution and batch size.
I use LMStudio for running models locally (macOS) and it tries to estimate whether the model would fit in my GPU memory (which is the same thing as main memory for Macs).<p>The Q4_K_S quantized version of Microsoft Fara 7B is a 5.8GB download. I'm pretty sure it would work on a 12GB Nvidia card. Even the Q8 one (9.5GB) could work.