Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    INDEX
    Explanations

    smoking

    np_max-act · gemini-2.0-flash

    # Neuron 4 Explanation The main thing this neuron does is find **quantitative comparisons and statistical differences between groups or conditions**. The neuron consistently activates on language describing measured outcomes, comparative results (higher/lower, increased/decreased, greater/less), statistical significance markers, and findings that show differences between groups—particularly in medical, scientific, and analytical contexts where results are being compared across conditions, demographics, or time periods.

    oai_token-act-pair · claude-4-5-haikuTriggered by @bowenyi

    passages reporting data-driven comparisons and statistical results (changes in risk or rates, quantified effects, and significance).

    oai_token-act-pair · gpt-5Triggered by @bowenyi
    New Auto-Interp
    Negative Logits
    _ly
    -0.07
     stressed
    -0.06
    _scaled
    -0.06
     poté
    -0.06
     Πά
    -0.06
    .readline
    -0.06
     Πα
    -0.06
     sıras
    -0.06
    	right
    -0.06
     오
    -0.06
    POSITIVE LOGITS
     Franken
    0.06
    ARP
    0.06
    .Disclaimer
    0.06
     аналог
    0.06
     cứu
    0.06
    :UIControlState
    0.06
     culprit
    0.06
     bubb
    0.06
    uner
    0.05
    ivet
    0.05
    Act Density 0.182%

    No Known Activations