© 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
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 28440
    Prev
    Next
    INDEX
    Explanations

    legal, scientific, or descriptive phrases

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

    The neuron’s activation rises steadily over the course of a passage, i.e. it detects how far “into” a segment a token appears, with later tokens triggering stronger activations.

    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
     before
    -1.51
     when
    -1.50
     because
    -1.49
     which
    -1.48
     will
    -1.43
     with
    -1.42
     at
    -1.40
     especially
    -1.38
     after
    -1.37
     and
    -1.33
    POSITIVE LOGITS
    perhaps
    1.69
    ーーーー
    1.52
     specif
    1.30
     flexibel
    1.30
     ฟ
    1.30
     VERY
    1.29
     контек
    1.29
     ambiti
    1.28
     colossal
    1.27
    certainly
    1.27
    Activations Density 0.008%

    No Known Activations