© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 14551
    Prev
    Next
    INDEX
    Explanations

    documentary context

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

    The neuron strongly activates on “technical” tokens—things like code‐style fragments (e.g. function names like pack), URL pieces (com, edit), hyphenated bits (on-), short symbol chunks (re, en), and other non-running-text tokens.

    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
     eenig
    -0.96
     eenige
    -0.95
    ogra
    -0.88
     Также
    -0.87
    }},
    -0.86
    in
    -0.85
    ville
    -0.84
    vore
    -0.84
    man
    -0.84
     vork
    -0.84
    POSITIVE LOGITS
     our
    1.02
     nové
    0.94
     these
    0.93
    ifflin
    0.91
     camisas
    0.91
     conversar
    0.90
     silueta
    0.90
    ないように
    0.90
    €¢
    0.90
     companies
    0.88
    Activations Density 0.004%

    No Known Activations