© 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. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-2B
    4. Residual Stream - 16k
    5. 18-GEMMASCOPE-RES-16K
    6. 4191
    Prev
    Next
    INDEX
    Explanations

    references to hearing or listening to others

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    "hear" or its inflections

    np_acts-logits-general · gemini-2.0-flash

    hearing and having heard

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.18.hook_resid_post
    Hook Layer
    18
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    SharedDtor
    -0.72
    UnknownFieldSet
    -0.62
     ModelExpression
    -0.59
     Picchu
    -0.57
    jspb
    -0.57
    ništvo
    -0.57
     Sols
    -0.57
    显
    -0.56
    martre
    -0.55
     uda
    -0.55
    POSITIVE LOGITS
     hear
    1.09
     hears
    1.08
     Hear
    0.98
     HEAR
    0.92
    Hear
    0.91
     heard
    0.91
    👂
    0.85
     voices
    0.82
    Heard
    0.81
     Heard
    0.81
    Activations Density 0.065%

    No Known Activations