© 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. 10-GEMMASCOPE-RES-131K
    7. 8348
    Prev
    Next
    INDEX
    Explanations

    hat, tha, tht

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

    The neuron consistently fires on the substring “that” (in its various sub-token forms).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
     or
    -2.63
    );
    -2.50
     and
    -2.31
     Because
    -2.23
    <td>
    -2.22
     of
    -2.08
    が
    -2.08
     We
    -2.06
    歲
    -2.05
    岁
    -2.03
    POSITIVE LOGITS
    
    2.50
    眛
    2.50
     they
    2.47
    祤
    2.45
    ſſen
    2.44
    𖤍
    2.41
    竝
    2.39
    ⩴
    2.25
    嘥
    2.23
     fateful
    2.22
    Activations Density 0.009%

    No Known Activations