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-2-2B
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 9400
    Prev
    Next
    INDEX
    Explanations

    important actions and events in a narrative context, particularly in sports reporting.

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

    descriptions of soccer moves and player actions in a match.

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot

    verbs and phrases related to scoring or achieving in sports and competitions. It activates strongly for words like "scored", "opened", "doubled", "netting", "winning goal", and similar phrases that indicate successful actions or outcomes in sports contexts.

    oai_token-act-pair · claude-3-5-sonnet-20240620Triggered by @bot

    Action verbs

    np_acts-logits-general · gemini-2.0-flash

    scored, missed, cut, netting

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71
    Prompts (Dashboard)
    36,864 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
    Cllr
    -0.82
     Roskov
    -0.70
     AssemblyCulture
    -0.69
     Cllr
    -0.67
    tonsoft
    -0.66
    Councillor
    -0.60
     Herrn
    -0.57
     congressman
    -0.56
    AnimationsModule
    -0.56
     Bürgermeister
    -0.55
    POSITIVE LOGITS
     pihaknya
    0.76
     society
    0.59
    咱们
    0.52
     החברה
    0.51
     our
    0.51
     Society
    0.51
    咱
    0.51
     us
    0.51
    TagMode
    0.50
     company
    0.50
    Activations Density 0.607%

    No Known Activations