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. 4661
    Prev
    Next
    INDEX
    Explanations

    specific names, particularly of people in political contexts.

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

    descriptions related to recent news and geopolitical events.

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

    proper names and titles, especially those related to government officials, politicians, and public figures.

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

    Follows titles, like "Adviser", "Secretary"

    np_acts-logits-general · gemini-2.0-flash

    titles followed by names

    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
     AnyObject
    -0.62
    Πηγές
    -0.60
    lusconi
    -0.58
     wearable
    -0.54
    Referanser
    -0.53
    anmoins
    -0.53
     libero
    -0.53
    jší
    -0.52
    KindOfClass
    -0.52
    خواندن
    -0.51
    POSITIVE LOGITS
    unknownFields
    0.69
     Coronavirus
    0.69
     Trump
    0.69
     coronavirus
    0.67
    coronavirus
    0.65
     useStyles
    0.64
    ExtendWith
    0.63
    Trump
    0.62
    Coronavirus
    0.62
     nahilalakip
    0.61
    Activations Density 0.155%

    No Known Activations