© 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
    3. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-9B
    7. Residual Stream - 16k
    9. 20-GEMMASCOPE-RES-16K
    11. 8921
    Prev
    Next
    INDEX
    Explanations

    constructs related to spatial or structural relationships

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

    endpoints of physical or conceptual ranges, particularly detecting phrases like "top and bottom", "beginning and end", or "starting and ending".

    oai_token-act-pair · claude-3-7-sonnet-20250219Triggered by @neilrathi

    references to boundaries or endpoints, especially paired “both ends”/“beginning and end” structures and items flanked or on each side.

    oai_token-act-pair · gpt-5Triggered by @jyhe0408

    words or tokens that mark boundaries, transitions, or endpoints in sequences, such as "end," "bottom," "trailing," or structural delimiters in text.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.20.hook_resid_post
    Hook Layer
    20
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ⓧ
    -0.48
    ThroughAttribute
    -0.44
    zi
    -0.41
     mijne
    -0.41
     second
    -0.39
     Atomic
    -0.39
     latter
    -0.38
    دانشنامهٔ
    -0.38
     atomic
    -0.38
     gatti
    -0.38
    POSITIVE LOGITS
    DockStyle
    0.58
    sidemargin
    0.50
     estekak
    0.50
    前后
    0.47
     obě
    0.47
     nahilalakip
    0.45
     flanking
    0.45
     extremes
    0.44
    ArgsConstructor
    0.43
     Sides
    0.43
    Activations Density 0.337%

    No Known Activations