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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B-IT
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 1107
    Prev
    Next
    INDEX
    Explanations

    modal verbs and phrases expressing obligations or possibilities

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

    Preceding modal verbs (would, can, must, should, will)

    np_acts-logits-general · gemini-2.0-flash

    modal verbs and subsequent actions

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91
    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
     quelque
    -0.31
     gouvern
    -0.27
     derni
    -0.26
    Personendaten
    -0.24
     vieill
    -0.24
     compañ
    -0.24
    ExecuteReader
    -0.23
     Mais
    -0.23
     aujourd
    -0.23
     nouve
    -0.23
    POSITIVE LOGITS
    ftagPool
    0.71
     use
    0.67
    OwnProperty
    0.66
     好文分享
    0.65
     utilize
    0.65
     give
    0.65
     Administrativna
    0.64
    Diwedd
    0.64
    <pad>
    0.63
    [@BOS@]
    0.63
    Activations Density 0.579%

    No Known Activations