© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 99902
    Prev
    Next
    INDEX
    Explanations

    profanity, curse words

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

    The neuron is looking for emphatic or “shouty” expletives and imperative‐style words (i.e. strong commands and profane intensifiers).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
     LOTS
    -0.92
    ijakan
    -0.90
    atisch
    -0.84
    нодоро
    -0.84
    こんな感じ
    -0.83
    anjutnya
    -0.82
    }/>
    -0.82
    Cantidad
    -0.81
    PARED
    -0.81
     tikra
    -0.81
    POSITIVE LOGITS
     fucking
    4.66
     damn
    4.06
     goddamn
    3.91
     freaking
    3.61
     freakin
    3.22
     damned
    3.19
     fuckin
    3.16
     frig
    3.06
    fucking
    3.02
     bloody
    2.84
    Activations Density 0.090%

    No Known Activations