News for hackers. Stuff that matters.
Re:Fmllm: 4mb training data, 100mb model, Fibonacci embeddin... by relaxing on Friday August 22, 2025 @09:09AM (#44984216)
I’d like to see the experiment run with other intervals besides the fibonacci sequence, and see if the results are any less coherent.
Re:Fmllm: 4mb training data, 100mb model, Fibonacci embeddin... by gabriel666smith on Friday August 22, 2025 @09:14AM (#44984272)
For interim adjacent results, I published https://github.com/henrygabriels/FMLLM/blob/main/improving_l... also, but I'm this moment working on a similar suggestion (comparing coherence 'fib-associated' and just 'random' words). Any intervals that you'd suggest?
Re:Fmllm: 4mb training data, 100mb model, Fibonacci embeddin... by relaxing on Friday August 22, 2025 @01:37PM (#44987335)
definitely random

logarithmic

primes

multiples of ten (or some other factor)

square numbers

Re:Fmllm: 4mb training data, 100mb model, Fibonacci embeddin... by gabriel666smith on Friday August 22, 2025 @01:43PM (#44987395)
On it! Thanks! Will be fun to learn how to do 'definitely random'.