© 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-3-12B
    3. 31-GEMMASCOPE-2-RES-16K
    4. 438
    Prev
    Next
    INDEX
    Explanations

    This neuron strongly activates on question cues—especially “wh-” question words (e.g. What, How) and function words that form questions (e.g. is) indicating a query.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_31_width_16k_l0_medium
    Prompts (Dashboard)
    392,802 prompts, 256 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    <0x0D>
    0.37
    เนื่อง
    0.35
    0
    0.32
    1
    0.32
                                   
    0.31
    }
    0.31
     เนื่องจาก
    0.30
                  
    0.30
    ]);
    0.29
    2
    0.29
    POSITIVE LOGITS
     they
    0.51
     it
    0.51
     THEY
    0.49
     does
    0.48
     you
    0.44
     exactly
    0.44
    soever
    0.44
     we
    0.43
     did
    0.42
     YOU
    0.42
    Activations Density 0.921%

    No Known Activations