© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 3576
    Prev
    Next
    INDEX
    Explanations

    The neuron activates on numeric expressions (figures, statistics, percentages, case or stock numbers) indicating quantities or measurements.

    oai_token-act-pair · o4-mini
    Triggered by @jyhe0408

    transitional phrases or conjunctions that connect clauses or sentences, particularly in formal or technical writing.

    oai_token-act-pair · claude-4-5-sonnet
    Triggered by @jyhe0408

    sentence-initial demonstrative cues that refer back to prior content and introduce a key result, conclusion, or emphasis in formal/news-style prose.

    oai_token-act-pair · gpt-5
    Triggered 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
    Negative Logits
     nikdy
    0.55
     sempre
    0.54
    經常
    0.54
    常に
    0.54
     alltid
    0.52
     supaya
    0.52
     usadas
    0.52
     makhluk
    0.52
     usado
    0.51
    デニム
    0.51
    POSITIVE LOGITS
     confirms
    1.41
     reinforces
    1.38
     reaff
    1.28
     coincides
    1.17
     represents
    1.16
     reaffirm
    1.16
     follows
    1.15
     joins
    1.15
    represents
    1.14
     exemplifies
    1.13
    Activations Density 0.368%

    No Known Activations