© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-2-27B
    3. 34-GEMMASCOPE-RES-131K
    4. 2118
    Prev
    Next
    INDEX
    Explanations

    as for/to

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

    The neuron strongly activates on standalone topic‐label nouns (e.g. “sexual,” “violence,” “policy,” “course,” “session,” “id,” “input,” “output”), effectively spotting section headings or formal content tags.

    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
     and
    -3.73
     but
    -2.63
    But
    -1.55
    but
    -1.50
     to
    -1.45
     BUT
    -1.44
     But
    -1.41
     along
    -1.34
     plus
    -1.27
     поэтому
    -1.25
    POSITIVE LOGITS
     there
    2.44
     it
    2.05
    については
    1.55
    而言
    1.53
     we
    1.47
     thì
    1.41
    に関しては
    1.41
     mantém
    1.40
     houve
    1.39
     platí
    1.38
    Activations Density 0.048%

    No Known Activations