© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-2-2B
    3. 0-GEMMASCOPE-TRANSCODER-16K
    4. 12782
    Prev
    Next
    INDEX
    Explanations

    the word "caught." It also seems to activate on the word "catch" to a lesser extent along with loosely associating it with related concepts such as mountain and reality.

    oai_token-act-pair · gemini-2.0-flash

    caught

    np_max-act · gemini-2.0-flash
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2b-pt-transcoders/layer_0/width_16k/average_l0_76
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.0.ln2.hook_normalized
    Architecture
    jumprelu_transcoder
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    Xna
    -0.64
    BindingResult
    -0.60
    ↵↵
    -0.60
    nestjs
    -0.59
     HttpHeaders
    -0.59
     ans
    -0.58
     pas
    -0.57
     contro
    -0.56
    rxjs
    -0.56
     so
    -0.55
    POSITIVE LOGITS
     avoient
    1.27
     étoient
    1.27
     auroit
    1.09
     mijne
    1.09
     feroit
    1.06
     ainfi
    1.05
     réfrig
    1.04
    seamnă
    1.02
     firent
    1.02
     aveug
    1.02
    Activations Density 1.699%

    No Known Activations