© 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
    3. Gemma-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 56091
    Prev
    Next
    INDEX
    Explanations

    huh or eh

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

    This neuron specializes in spotting software‐help and troubleshooting language—brand or product names (like Norton), user‐directed terms (“you,” “can”), and program-related words.

    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
     τά
    -0.79
    expression
    -0.79
    なんです
    -0.74
    なんだけど
    -0.74
    Knox
    -0.74
    drv
    -0.72
     때문에
    -0.72
     minutes
    -0.72
    etas
    -0.72
     bucket
    -0.72
    POSITIVE LOGITS
     huh
    5.94
     eh
    4.94
    huh
    3.91
     Huh
    3.69
    Huh
    3.44
    eh
    3.23
     Eh
    3.11
    Eh
    2.91
     EH
    2.19
    EH
    2.13
    Activations Density 0.050%

    No Known Activations