© 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
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 72030
    Prev
    Next
    INDEX
    Explanations

    introducing contrast

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

    The neuron detects contrastive discourse markers—primarily words like “but” and “however” that signal a shift or contrast in the text.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/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
     också
    -1.23
     également
    -1.17
    也是
    -1.16
     також
    -1.16
     også
    -1.16
     keramik
    -1.09
     also
    -1.06
     STRATEG
    -1.06
    首先
    -1.06
     също
    -1.05
    POSITIVE LOGITS
     зато
    1.39
    倒是
    1.23
    ناك
    1.11
     immerhin
    1.11
     several
    1.08
     plenty
    1.08
     other
    1.07
     later
    1.07
     two
    1.05
     through
    1.02
    Activations Density 0.036%

    No Known Activations