You know the LLM is just something that's trained on human tokens. We can probab...

You know the LLM is just something that's trained on human tokens.

We can probably create a dog intelligence by training it on dog tokens. Barks and stuff.

Same with dolphins. I wonder if multimodal models that know english tokens and dolphin tokens can cross the gap? Something to experiment with.

First sentence we would understand from Dolphins language: thanks for all the fish!