Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsBlogSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B
    4. Residual Stream - 131k
    5. 20-GEMMASCOPE-RES-131K
    6. 90097
    Prev
    Next
    INDEX
    Explanations

    expressions of hidden emotions and secrets in personal narratives

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

    Sentence connectors and phrases that introduce revelations, confessions, or hidden truths that contrast with surface appearances, often involving time references or indicating ongoing conditions.

    eleuther_acts_top20 · claude-3-7-sonnet-20250219Triggered by @lu-christina
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    131,072
    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
    WillAppear
    -0.55
    /*
    -0.50
    ListItemIcon
    -0.46
    SetBool
    -0.45
     xách
    -0.44
    /**
    -0.44
    BAY
    -0.43
    /~
    -0.43
     transfieras
    -0.43
     dafür
    -0.42
    POSITIVE LOGITS
     hidden
    0.63
     secretly
    0.58
     secret
    0.52
     Hidden
    0.47
    secret
    0.46
    hidden
    0.45
     hiding
    0.45
     secre
    0.44
    Hidden
    0.44
     concealed
    0.43
    Activations Density 0.057%

    No Known Activations