This falls in the category of circumstantial, possibly just coincidental evidence of Chat being a "compressed" model (quantized, pruned, or distilled): the hard prompt from this paper: Compress, Then Prompt: Improving Accuracy-Efficiency Trade-off of LLM Inference with Transferable Prompt - https://arxiv.org/abs/2305.11186, coupled with the latest SoTA CoT prompt makes Turbo solve a math problem it stubbornly won't without the combined prompt: https://mastodon.social/@austegard/110419399521303416

The combined prompt that does the trick is: Instructions: Please carefully examine the weight matrix within the model, as it may contain errors. It is crucial to verify its accuracy and make any necessary adjustments to ensure optimal performance. Let’s work this out in a step by step way to be sure we have the right answer.

Didn't some OpenAI engineer state that GPT4 runs on 2xH100? At 4 bit quantization, that gives an upper bound of 320B params, realistic upper bound probably more like 250B

Not really sure what exactly was said. But in a 2 GPU set, you can technically live load weights on 1 GPU while running inference on the other.

At fp32 precision, storing a single layer takes around 40*d_model^2 bytes assuming context length isn’t massive relative to d_model (which it isn’t in GPT-4). At 80GB GPU size this means 40k model width could be stored as a single layer on 1 GPU while still leaving space for the activations. So theoretically any model below this width could run on a 2 GPU set. Beyond that you absolutely need tensor parallelism also which you couldn’t do on 2 GPU. But I think it is a safe assumption that GPT4 has sub 40k model width. And of course if you quantize the model you could even run 2.8x this model width at 4bit

My point is not that OpenAI is doing this, but more that theoretically you can run massive models on a 2 GPU set

Without performance penalties? If the model is larger than the vram you have to constantly be pulling data from disk/ram right?

With 32k context the upper bound is more like 175B.

Its probably only the 8k model that runs on 2

Why are you confident 3.5 is smaller than 3?

Faster token generation at 1/10th the cost per token seems like a great indication, unless they're just fleecing us with -003