© 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. 320
    Prev
    Next
    INDEX
    Explanations

    conditional statements

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

    The neuron is highly activated by numeric literals (e.g. “270”, “266”, etc.), so it’s essentially detecting number tokens in code.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    blank lines or whitespace in code.

    oai_token-act-pair · claude-4-5-haikuTriggered 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
     muñeco
    -1.12
     zarówno
    -1.07
    Itr
    -1.06
    の準備
    -1.05
     gamla
    -1.05
     első
    -1.05
     både
    -1.05
     VARIOUS
    -1.04
    patience
    -1.04
     protéger
    -1.02
    POSITIVE LOGITS
     while
    1.19
     if
    1.17
     since
    1.15
     when
    1.05
    ínico
    1.03
     some
    0.97
     потому
    0.96
    Напомним
    0.96
    boldsymbol
    0.92
     but
    0.90
    Activations Density 0.002%

    No Known Activations