Hugging Face's logo
Jonathan Harrison's picture
12 2 5

Jonathan Harrison PRO

Raiff1982
·
https://www.raiffsbits.com

AI & ML interests

i want to learn with it

Recent Activity

replied to grimjim's post about 1 hour ago
This recent paper points to an explanation for the unreasonable effectiveness of Frankenmerges: https://huggingface.co./papers/2502.05171 Specifically, the duplication of layers in Frankenmerges serves a purpose similar to what occurs in their recurrent-depth architecture. Successful frankenmerges that operate without additional fine-tuning are able to recover or "heal" from any damage due to abrupt transitions between layer blocks. Operational replicated layer blocks can provide functional benefits grounded in latent reasoning. Frankenmerges can also result in hybrid reasoning, by splicing together the latent reasoning of different models. Back in April 2024, I was able to duplicate a few layers in the Llama 3 8B model, turning it into a 9B model, without harming benchmarks significantly, despite any transition damage. https://huggingface.co./grimjim/llama-3-experiment-v1-9B My informal experimentation suggested that latent reasoning circuits could occupy continguous stacks of 2-4 layers, though the result was highly sensitive to the choice of transition location between layers.
updated a Space about 21 hours ago
RaiffsBits/README
published a model about 21 hours ago
RaiffsBits/deep_thought
View all activity

Organizations

Raiff's Bits LLC's profile picture

Collections 1