[Proposal] Support SmolLM2

[neverix]

Proposal

Support the SmolLM2 series of models - https://huggingface.co/HuggingFaceTB/SmolLM2-135M etc

Motivation

SmolLM2 are models with the Llama architecture that are trained on better data than older models of the same size, especially the 135M variant. It has the minor weakness of having a broken 11th layer, but it is still becoming popular.

Pitch

Add the name/alias for SmolLM2 to the model list.

Checklist

• I have checked that there is no similar issue in the repo (required)