Nvidia releases NVLM 1.0 72B open weight model

Not a lawyer, not legal advice, but... the legal status quo is that neural network outputs are not copyrightable. They are currently considered not made by humans nor considered a derivative work from the training material / network weights (assuming it's not regurgitating copyrighted material verbatim).

The cc-by-nc-4.0 license applies to the network weights. The only thing non-commercial about the license is that it restricts how you may reproduce the licensed material:

> reproduce and Share the Licensed Material, in whole or in part, for NonCommercial purposes only; and

As long as you are not selling the network weights themselves, nothing in the license prevents you from evaluating the neural network for commercial purposes and selling the outputs. In 'production' you will have to directly download the weights from Nvidia themselves (or another 3rd party which is distributing the network weights non-commercially in good faith) though, you can't share the network weights onto your commercial inference server from another one of your commercial deployment servers. Or at least, it gets more dicy there and may be considered commercial reproduction so better avoid it.

For similar reasons you may 3D print a CC-BY-NC model of a tool and use that tool in your commercial workshop, you may use a CC-BY-NC compiler of a language to compile commercial programs, etc.

I think the answer here is "it depends." The Llama-3.2 series is an extended version of the Llama-3.1 series with multimodal (image) training, but they kept the language model weights frozen and only updated the new image weights. So in the end, the 3.2 series benchmarks identically to 3.1 on text-only tasks; the image weights provided no value to the language model weights.

Allowing the language model weights to be updated during training could potentially result in better performance on both tasks, though, if Nvidia's result replicates. I could believe that it might: after all, more diverse data is more diverse data, and the model will be forced during training to generalize more.

Llama-3-V models do that, but are not published.

That's natural given that they mostly produce hardware several layers of abstraction distant from the end user value, companies need to buy the hardware before they can start delivering their own value. AI model training is not value by itself if there's no use-case for the model that can be charged for.

I see it playing out one of two ways. Either Nvidia are selling shovels in a gold rush, the rush will end, and the business will dry up (after they have made a lot of money!). Or AI sticks/takes off, and Nvidia are selling a commodity too far from the value, like most electronic component manufacturers, and they'll maintain significant market share but have their margins reduced to a fraction of what they were before (after they made a lot of money!).

The human value doesn't come from ML training or inference, it comes from taking a better photo. The business value comes from drafting a better email. Those companies closer to that value will likely do better in the long run, as they always have done.

Wrong

Midjourney is profitable. All the acquired startups (i.e. Streamlit or MosaicML) who made millions per employee "made money" for the people who cared.

I'm pretty sure https://www.topazlabs.com/ is also making money with the AI revolution.

Also Klarna threw out 700 people, they probably make money with AI.

And i found this article: https://www.ft.com/content/a9a192e3-bfbc-461e-a4f3-112e63d0b...

Its an revolution. Don't undersell this.

There was never ever any technology like LLMs close to what chatgpt and co can do in regards of understanding random human input.

My startup doesn't need to make money with it directly, but for us it increased our data quality on text and images.

I'm also quite happy to pay 10-20$ per month for random things LLMs do quite well for different use cases like creating some scripts etc.

That's not true, there are plenty of companies that make a profit, Midjourney, for example, an obvious one.

"When there is a gold rush, sell shovels"

i have yet to hear of anyone actually using AI for something properly

only exception im excited about is the non-main characters from video games, where a lot of the random NPCs, can now actually bring some more fun to the game.

144GB VRAM to load the weights at FP16, 72GB quantized to FP8. To figure out the KV cache size you'll need for an LLM, you can use the following formula: https://x.com/AlpinDale/status/1841305040545329535

Simplified for posterity:

    kv_bytes = kv_bits / 8
    hidden_per_head = hidden_size // num_attention_heads
    total_heads = hidden_per_head * num_key_value_heads
    kv_bytes_per_token = 2 * kv_bytes * num_hidden_layers * total_heads

(Edit: I accidentally swapped in some of the vision config bytes in my original calculation; these are the corrected numbers.) So, for NVLM 1.0 72B, that works out to 640kb per token assuming FP16 KV cache. If you use the entire 32k context length, that's an extra ~20GB of overhead for the KV cache. Then depending on how you're running the LLM, there might be extra overhead e.g. compiled CUDA graphs.

You can cut this down lower by using grouped query attention as described here: https://medium.com/@plienhar/llm-inference-series-4-kv-cachi... This allows you to divide that number by the number of grouped heads, although it trades off accuracy for VRAM usage.

But TLDR, a minimum of around 164GB of VRAM at full accuracy. To me that seems fairly low, and I think vLLM would OOM without significantly more than that, but that's about as low as you could go in theory if you're running everything at FP16. Half that, of course, for FP8.

You'll typically need to have a copy of the KV cache per GPU, if you're using multiple GPUs, so multiply the KV cache overhead by the number of GPUs you're using. This will depend on what the specs for the GPUs you're using are; for example, you'll need 3 H100s (really four, since vLLM wants the number of heads to be evenly divisible by the number of GPUs); if you're using L40Ses, you'll need eight of them; but most likely only a single AMD MI300x.

I haven't tested it, but likely around 170GB, regardless of if you're using only one GPU or spreading it out over several ones.

Am I looking at the wrong table? It dominates everything on visual interpretation benchmarks.

Edit: specifically ocrbench and VQAv2

It's not that bad, and I'd much rather that they be honest instead of lying like everyone else does.

All jokes aside (and that did make me laugh) at least they're not training just to hit the benchmarks, which seem to be more meaningless as a quality indicator with each passing day.

I see at a few models (3 models in MMMU) that score lower than Nvidia's. But putting that aside, they at least get points for apparent objectivity. At least they probably aren't fudging numbers.

Well but it actually doesn't, unless you're looking only at MMMU.