Multimodal Large Language Model

why not LLMM?

Because the first word is "multimodal" :^) and also because MLLM is the established initialism.

My point was the phrase already contains the word model. Why are we calling it a multimodel LL model? Why not just add multi to the existing model?

Multimodal, not multimodel. Multimodal referring to the different possible modes of input (text, picture) into the model.

There is something like a "multimodel LLM" but it's called MoE ("mixture of experts").

Modal, not model

Modalities and models are not the same thing.

MultimodAl, not multimodEl

what's a language multimodal model

Ok our options

  Multimodal large language model
  Large multimodal language model
  Large language multimodal model
  Large language model (multimodal)

I prefer 1, because this is a multimodal type of an existing technique already referred to as LLM. If I was king, I’d do Omnimodal Linguistic Minds, but no one asks me such things, thank god

I mean, if we want to be silly, what about "Language model (large, multimodal"? :)

Some of the modalities in multimodal are non-linguistic. For example, image or video input. In those cases, is it still a language model?

I’d say yeah cause it’s understanding the inputs through linguistic structures/patterns

The bikeshed color argument never ceases to be relevant. Would you say "large language model multimodal"? I doubt it.

Just a thought not a bike shed, relax.

Also, is FERRET an acronym?

I would guess it’s wordplay on other models being named after animals (llama, vicuña) and figurative use of “ferret”.

https://en.m.wiktionary.org/wiki/ferret: “3. (figurative) A diligent searcher”

The language model works by delegating tasks to smaller language models and overcharging them for GPU time.