© 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. 4158
    Prev
    Next
    INDEX
    Explanations

    negative or critical sentiments

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

    references to societal discussions, debates, or collaborative projects examining complex topics.

    oai_token-act-pair · claude-3-5-haiku-20241022Triggered by @neilrathi

    end-of-sentence or clause boundaries, with strongest activation on final punctuation and closing punctuation marks.

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

    sentence-ending periods, particularly in formal or technical writing.

    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
    EDEFAULT
    -0.76
    RegressionTest
    -0.71
     informée
    -0.68
     autorytatywna
    -0.65
     oprot
    -0.64
     IBOutlet
    -0.63
     विश्वसनीयता
    -0.63
    Personensuche
    -0.62
     AssemblyCulture
    -0.61
    izedBox
    -0.60
    POSITIVE LOGITS
    enumi
    0.49
     indépend
    0.35
     hoja
    0.33
     subjects
    0.33
     aims
    0.31
    Dr
    0.31
    ({})
    0.30
     clave
    0.29
     interesa
    0.29
    using
    0.28
    Activations Density 0.394%

    No Known Activations