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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-2-27B
    3. 10-GEMMASCOPE-RES-131K
    4. 99408
    Prev
    Next
    INDEX
    Explanations

    truth be told

    np_acts-logits-general · gemini-2.5-flash-lite

    The neuron responds most strongly to very frequent English function words (common stopwords like be, as, or, in, for, and, at, etc.).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    common function words and grammatical connectors (prepositions, conjunctions, articles, and auxiliary verbs) that structure sentences.

    oai_token-act-pair · gpt-5Triggered by @jyhe0408

    common idiomatic phrases and fixed expressions (like "truth be told", "at least in part", "more-or-less", "and of themselves", "for sure", "of course").

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     it
    -2.52
    7
    -1.78
     there
    -1.59
     schermata
    -1.45
     menyadari
    -1.44
     the
    -1.40
     by
    -1.40
    parently
    -1.39
    3
    -1.36
    9
    -1.36
    POSITIVE LOGITS
    </em>
    1.82
    /}{
    1.70
    تیجه
    1.70
    $\$
    1.61
     }
    1.59
    </td>
    1.57
    </sup>
    1.57
    secutions
    1.55
    },
    
    1.50
    $,
    1.49
    Activations Density 0.446%

    No Known Activations