© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-3-12B-IT
    3. 12-GEMMASCOPE-2-RES-262K
    4. 38570
    Prev
    Next
    INDEX
    Explanations

    The task is to explain the behavior of a neuron concisely using the provided lists.The explanation should be a phrase between 3 and 20 words, describing what the neuron detects or predicts by finding patterns.Let's look at the lists:* **MAX\_ACTIVATING\_TOKENS**: `Mac`, `people`, `Mac`, `Mac`, `NAS`, `Red`, `[*`, `people`, `and`, `Specifically` * Recurring tokens: `Mac`, `people`. * Other specific tokens: `NAS`, `Red`, `Specifically`.* **TOKENS\_AFTER\_MAX\_ACTIVATING\_TOKEN**: `).`, `-`, `)`, `).`, `(`, `es`, `].`, `they`, `analyzing`, `,` * These are mostly punctuation or short suffixes/prefixes, suggesting grammatical context or list markers.* **TOP\_POSITIVE\_LOGITS**: `だって`, `с`, `ع`, `ட`, `其他`, `社会`, `した`, `ih`, `お`, `íamos` * This list is dominated by tokens in non-English languages, including Japanese, Cyrillic, Arabic, Tamil, Chinese, and Portuguese.**multilingual context or specific items like mac/nas/red**

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-it/resid_post/layer_12_width_262k_l0_medium
    Prompts (Dashboard)
    238,145 prompts, 512 tokens each
    Dataset (Dashboard)
    lmsys + oasst1
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     यामुळे
    1.01
     whatnot
    0.97
     ellipso
    0.94
     prioritized
    0.92
     ferrite
    0.91
     contaminate
    0.90
     automate
    0.89
     incub
    0.89
     interferon
    0.89
     gradually
    0.88
    POSITIVE LOGITS
    だって
    1.05
    с
    1.04
    ع
    1.02
    ட
    1.02
    其他
    0.96
    社会
    0.96
    した
    0.93
    ih
    0.93
    お
    0.93
    íamos
    0.93
    Activations Density 0.001%

    No Known Activations