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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-2-27B
    3. 10-GEMMASCOPE-RES-131K
    4. 111224
    Prev
    Next
    INDEX
    Explanations

    Erlang, names, languages

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

    proper nouns that are uncommon or technical terms, particularly names of programming languages, software, places, and people.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @bfgcn7frf5

    The neuron fires strongly on multi‐letter names or technical terms (e.g. Cuckoo, Erlang, Likert, Manteca), i.e. proper or domain-specific nouns.

    oai_token-act-pair · o4-miniTriggered 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
    </strong>
    -2.25
     spezi
    -1.79
    粜
    -1.77
     ")[
    -1.75
    </em>
    -1.73
     oscuro
    -1.72
     🔥
    -1.71
    didn
    -1.70
    が無
    -1.70
    honestly
    -1.66
    POSITIVE LOGITS
     When
    2.19
     From
    1.91
    
    1.89
    鴣
    1.85
     While
    1.85
     Rezept
    1.75
     Because
    1.68
     Having
    1.67
    when
    1.67
     Some
    1.67
    Activations Density 0.061%

    No Known Activations