Latent Lattice Language Models

(shivamduggal4.github.io)

11 points | by E-Reverance 2 hours ago ago

3 comments

  • augment_me an hour ago ago

    Seems like just an extension of Short-conv in the linear attention methods where we instead carry the EMA of the entire sequence instead of nearest 4 tokens.

    It would make sense in my opinion to do the evals based on the reason for needing Short-conv in the architectures that use it in the first place, where it was added to support exact retrieval/needle-in-haystack when we compress the attention state.

    It's odd to make this arch extension and not test beyond a 2k context window, or with the needle-in-haystack benchmark.

    In my head this should fail because the EMA cannot carry this granular information

  • royal__ 24 minutes ago ago

    How significant is a .98 decrease in perplexity. I'm a researcher in this field and I'm having trouble interpreting the implication of this.

  • tkslucas 19 minutes ago ago

    Interesting approach to scale compute, never thought of using the per layer state like that