© 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. 11201
    Prev
    Next
    INDEX
    Explanations

    articles and determiners in the text

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

    articles and other basic determiners marking the start of noun phrases.

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

    Looking at the activations, this neuron consistently fires strongly (values around 34-36) on the word "the" appearing in technical, academic, or formal writing contexts - particularly in scientific papers, legal documents, historical texts, and mathematical proofs. The neuron is detecting **

    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
     Autorisations
    -0.69
     EconPapers
    -0.62
    +#+
    -0.60
     Autorizaciones
    -0.58
    GOTREF
    -0.57
    IsContent
    -0.56
    SuspendLayout
    -0.56
     BoxDecoration
    -0.55
    IUrlHelper
    -0.52
    WireFormatLite
    -0.52
    POSITIVE LOGITS
    whose
    0.47
    ซึ่ง
    0.45
    which
    0.45
     mikä
    0.43
     cuyos
    0.43
     which
    0.43
    یکه
    0.43
     cuál
    0.42
     što
    0.42
     كومونز
    0.41
    Activations Density 0.187%

    No Known Activations