News for hackers. Stuff that matters.
Re:Show HN: Llama 3.2 Interpretability with Sparse Autoencoders by sinuhe69 on Friday November 22, 2024 @12:16AM (#42211372)
Not in math.
Re:Show HN: Llama 3.2 Interpretability with Sparse Autoencoders by TeMPOraL on Friday November 22, 2024 @01:12AM (#42211546)
Yes in math. Formalisms come after casual thoughts, at every step.
Re:Show HN: Llama 3.2 Interpretability with Sparse Autoencoders by sinuhe69 on Friday November 22, 2024 @04:04AM (#42212253)
What is a casual thought that you cannot explain in math?
Re:Show HN: Llama 3.2 Interpretability with Sparse Autoencoders by TeMPOraL on Friday November 22, 2024 @05:25AM (#42212620)
That question makes no sense. You can explain anything in math, because math is a language and lets you define whatever terms and axioms you need at a given moment.

(Whether or not such explanation is useful for anything is another issue entirely.)

Re:Show HN: Llama 3.2 Interpretability with Sparse Autoencoders by worldsayshi on Friday November 22, 2024 @06:34AM (#42212963)
Can you explain how intuition led you to try a certain approach?
Re:Show HN: Llama 3.2 Interpretability with Sparse Autoencoders by TeMPOraL on Friday November 22, 2024 @07:35AM (#42213323)
Is it enough if I hand-wave it with probability distributions, or do you want me to write out adjacency search in a high-dimensional space?
Re:Show HN: Llama 3.2 Interpretability with Sparse Autoencoders by mdp2021 on Friday November 22, 2024 @04:37AM (#42212427)
It's totally different: those formalisms are in a workbench, following a set of rules that either work or not.

So, yes, that (math) is representative of the actual process: pattern recognition gives you spontaneous ideas, that you assess for truthfulness in conscious acts of verification.

Re:Show HN: Llama 3.2 Interpretability with Sparse Autoencoders by legel on Friday November 22, 2024 @04:11PM (#42217404)
Math comes from brains.
Re:Show HN: Llama 3.2 Interpretability with Sparse Autoencoders by HeavyStorm on Sunday November 24, 2024 @11:15AM (#42228640)
That's some misunderstanding of the human brain and thought process...