© 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. 569
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on subjective stance markers and commentary words—e.g. personal‐opinion cues (I think, I know, I’m proud), discourse pivots (then, here, so, well), and other evaluative/adverbial tokens signaling the author’s attitude.

    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
     tzv
    0.52
     cosidd
    0.46
     spp
    0.44
     keltiradi
    0.43
     które
    0.42
     mathbf
    0.42
    /
    0.42
     rectangular
    0.41
     następnie
    0.41
     çalışmaları
    0.41
    POSITIVE LOGITS
    !
    0.83
    !?
    0.70
    !!
    0.66
    !...
    0.66
    !:
    0.65
    !!!
    0.63
     ;-)
    0.63
    ㅎㅎ
    0.63
    !'
    0.63
    !“
    0.63
    Activations Density 3.099%

    No Known Activations