© 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. 39871
    Prev
    Next
    INDEX
    Explanations

    explains why

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

    This neuron is primarily triggered by parenthetical punctuation characters (opening and closing parentheses, often around multiple‐choice labels).

    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
    assertThrows
    -0.93
    ratislava
    -0.93
    pielt
    -0.91
    volut
    -0.89
     Polonia
    -0.88
    desde
    -0.87
    failed
    -0.86
     séjours
    -0.86
    vania
    -0.85
     Kühlschrank
    -0.84
    POSITIVE LOGITS
     because
    3.67
     reasons
    2.59
     reason
    2.42
    because
    2.19
    เพราะ
    2.13
    Because
    2.08
     karena
    2.06
    的原因
    2.00
     Because
    1.99
     porque
    1.97
    Activations Density 0.069%

    No Known Activations