© 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-IT
    7. Residual Stream - 131k
    9. 31-GEMMASCOPE-RES-131K
    11. 41628
    Prev
    Next
    INDEX
    Explanations

    punctuation and formatting characters, particularly related to strings and delimiters in coding contexts

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

    delimiters like comma and semicolon

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 31-gemmascope-res-131k
    Configuration
    google/gemma-scope-9b-it-res/layer_31/width_131k/average_l0_109
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.31.hook_resid_post
    Hook Layer
    31
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ագրություններ
    -0.70
    RTEX
    -0.64
    ⟬
    -0.55
     pinulongan
    -0.54
    AddTagHelper
    -0.51
     esternos
    -0.50
    RefNanny
    -0.50
    mitives
    -0.49
     Signalez
    -0.49
     Taktlose
    -0.49
    POSITIVE LOGITS
     comma
    1.20
    comma
    0.96
     commas
    0.94
    Comma
    0.93
     coma
    0.92
     Com
    0.82
     semicolon
    0.79
     COM
    0.79
    逗
    0.73
     Coma
    0.71
    Activations Density 0.367%

    No Known Activations