Does anyone know whether the 128K is input tokens only? There are a lot of models that have a large context window for input but a small output context. If this actually has 128k tokens shared between input and output, that would be a game changer.
Mistral NeMo
141–150 of 165 posts
Re: Mistral NeMo
#142Earlier quoted context omitted.
I don't have explanations but I can point you to one of the papers: https://arxiv.org/pdf/2309.12288 which calls it "the reversal curse" and does a bunch of experiments showing models that are successful at questions like "Who is Tom Cruise’s mother?" (Mary Lee Pfeiffer) will not be equally successful at answering "Who is Mary Lee Pfeiffer’s son?"
Isn't that specific case just a matter of not having enough data _explicitly_ stating the reverse? Seems as if they are indeed stochastic parrots from that perspective.
Yes, and your conclusion is correct.
Re: Mistral NeMo
#143I have to say, the experience of trying to sign up for Nvidia Enterprise so you can try the "NIM" packaged version of this model, is just icky and and awful now that I've gotten used to actually free and open models and software. It feels much nicer and more free to be able to clone llama.cpp and wget a .gguf model file from huggingface without any registration at all. Especially since it has now been several hours since I signed up for the Nvidia account and it still says on the website "Your License Should be Active Momentarily | We're setting up your credentials to download NIMs."
I really don't get Nvidia's thinking with this. They basically have a hardware monopoly. I shelled out the $4,000 or so to buy two of their 4090 GPUs. Why are they still insisting on torturing me with jumping through these awful hoops? They should just be glad that they're winning and embrace freedom.
Re: Mistral NeMo
#144I just checked huggingface and the model files download is about 25GB but in a comment below someone mentioned it is 8fp quantized model. Trying to understand how the quantization affects the model (and RAM) size. Can someone please enlighten.
Re: Mistral NeMo
#145Earlier quoted context omitted.
isn't it 2 bytes (fp16) per param. so 7b = 14 GB+some for inference?
This was trained to be run at FP8 with no quality loss.
The model description on huggingface says - Model size - 12.2B params, Tensor type - BF16. Is the Tensor type different from the training param size?
Re: Mistral NeMo
#146I have to say, the experience of trying to sign up for Nvidia Enterprise so you can try the "NIM" packaged version of this model, is just icky and and awful now that I've gotten used to actually free and open models and software. It feels much nicer and more free to be able to clone llama.cpp and wget a .gguf model file from huggingface without any registration at all. Especially since it has now been several hours s…
This is what you get when managers design a software tool instead of engineers designing it.
Re: Mistral NeMo
#147> Today, we are excited to release Mistral NeMo, a 12B model built in collaboration with NVIDIA. Mistral NeMo offers a large context window of up to 128k tokens. Its reasoning, world knowledge, and coding accuracy are state-of-the-art in its size category. As it relies on standard architecture, Mistral NeMo is easy to use and a drop-in replacement in any system using Mistral 7B. > We have released pre-trained base an…
if you want to be lazy, 7b = 7gb of vRAM, 12b = 12gb of vRAM, but quantizing you might be able to do with with ~6-8. So any 16gb Macbook could run it (but not much else).
What about for fine-tuning? Are the memory requirements comparable to inference? If not, is there a rule of thumb for the difference? Would it be realistic to do it on a macbook with 96G of unified memory?
Re: Mistral NeMo
#148I just checked huggingface and the model files download is about 25GB but in a comment below someone mentioned it is 8fp quantized model. Trying to understand how the quantization affects the model (and RAM) size. Can someone please enlighten.
Sure. The talk about 8bit refers to quantization-aware training. Pretty common in image models these days to reduce the impact of quantization on accuracy.
Typically this might mean that you simulate an 8bit forward pass to ensure that the model is robust to quantization ‘noise’. You still use FP16/32 for backward pass & weight updates for numerical stability.
It’s just a way to optimize the model in anticipation of future quantization. The experience of using an 8-bit Nemo quant should more closely mirror that of using the full-fat bf16 model compared to if they hadn’t used QAT.
Re: Mistral NeMo
#149These big models are getting pumped out like crazy, that is the business of these companies. But basically, it feels like private/industry just figured out how to scale up a scalable process (deep learning), and it required not $M research grants but $BB "research grants"/funding, and the scaling laws seem to be fun to play with and tweak more interesting things out of these and find cool "emergent" behavior as billi…
I don’t see any indication this beats Llama3 70B, but still requires a beefy GPU, so I’m not sure the use case. I have an A6000 which I use for a lot of things, Mixtral was my go-to until Llama3, then I switched over. If you could run this on say, stock CPU that would increase the use cases dramatically, but if you still need a 4090 I’m either missing something or this is useless.
Comparing this to 70b doesn't make sense: this is a 12b model, which should easily fit on consumer GPUs. A 70b will have to be quantized to near-braindead to fit on a consumer GPU; 4bit is about as small as you can go without serious degradation, and 70b quantized to 4bit is still ~35GB before accounting for context space. Even a 4090 can't run a 70b.
Supposedly Mistral NeMo better than Llama-3-8b, which is the more apt comparison, although benchmarks usually don't tell the full story; we'll see how it does on the LMSYS Chatbot Arena leaderboards. The other (huge) advantage of Mistral NeMo over Llama-3-8b is the massive context window: 128k (and supposedly 1MM with RoPE scaling, according to their HF repo), vs 8k.
Also, this was trained with 8bit quantization awareness, so it should handle quantization better than the Llama 3 series in general, which will help more people be able to run it locally. You don't need a 4090.
Re: Mistral NeMo
#150Earlier quoted context omitted.
Justine Tunney (of redbean fame) is actively working on getting LLMs to run well on CPUs, where RAM is cheap. If successful this would eliminate an enormous bottleneck to running local models. If anyone can do this, she can. (And thank you to Mozilla for financially supporting her work). See https://justine.lol/matmul/ and https://github.com/mozilla-Ocho/llamafile
There's no good reason for consumer nvidia cards to lack SODIMM-like slots for video RAM, except to rake in big bucks and induce more hasty planned obsolescence.
DIMM slots won't work for GPU VRAM due to the higher speeds, tighter signalling, and dense packing of memory on wide buses. Take a look at the speeds DDR5 is running at in a typical Xeon server, and compare to GDDR6. This is the problem LPCAMM2 was developed to solve for modern x86 CPUs in laptops and desktops. Seeing it applied to GPUs would be great.