© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 8558
    Prev
    Next
    INDEX
    Explanations

    elements related to mathematical concepts and functions

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

    LaTeX math markup patterns, especially spacing/kerning commands and surrounding inline equation structure (e.g., \mkern, -2mu, \smash, \underline within equations and captions).

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

    LaTeX formatting code, particularly the `\mkern` command with negative spacing values.

    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
    LookAnd
    -0.79
    parsedMessage
    -0.77
    RegressionTest
    -0.75
     betweenstory
    -0.73
    GEBURTSDATUM
    -0.69
     CreateTagHelper
    -0.65
    msgTypes
    -0.57
    ⟬
    -0.56
    AutoScaleMode
    -0.55
    ViewImports
    -0.53
    POSITIVE LOGITS
    seamnă
    0.40
    quium
    0.35
     Obtenez
    0.34
    ępo
    0.34
    hæng
    0.33
     Raceway
    0.33
     FRUIT
    0.33
     symboles
    0.32
     Romains
    0.32
    ăsp
    0.32
    Activations Density 0.101%

    No Known Activations