© 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-2-9B-IT
    4. Residual Stream - 131k
    5. 20-GEMMASCOPE-RES-131K
    6. 94587
    Prev
    Next
    INDEX
    Explanations

    questions and references to events related to societal issues and human rights

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

    foreign language texts

    np_acts-logits-general · gemini-2.5-flash-lite

    This neuron detects potentially sensitive or policy-relevant tokens, such as place names, explicit sexual/violent terms, and other identifiable or flagged content.

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0

    common function words and punctuation that mark clause boundaries and basic connectors in sentences.

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k
    Configuration
    google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81
    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
     disambiguazione
    -0.51
    Хьажоргаш
    -0.50
    stdc
    -0.50
    TestTools
    -0.46
     otomatig
    -0.43
    ChildScrollView
    -0.42
    ViewFeatures
    -0.41
     واق
    -0.41
    ValueStyle
    -0.41
     nahilalakip
    -0.40
    POSITIVE LOGITS
     artistique
    0.47
    astéroïdes
    0.46
     AssemblyTitle
    0.44
     spirituale
    0.43
     artístico
    0.41
     desmotivaciones
    0.39
     banderas
    0.39
     artistico
    0.38
     palsu
    0.38
     vänner
    0.38
    Activations Density 0.050%

    No Known Activations