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. 10278
    Prev
    Next
    INDEX
    Explanations

    past tense verbs or phrases indicating actions taken or conditions existing

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

    Past participle form of "to be" (been, being, was, were, be)

    np_acts-logits-general · gemini-2.0-flash

    forms of 'to be' followed by past participle

    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
     réalis
    -0.86
     Jefus
    -0.73
     dieux
    -0.71
     purpoſe
    -0.71
     Chriſt
    -0.71
     ſche
    -0.70
     avoient
    -0.70
     ſeveral
    -0.68
    wrote
    -0.68
     ſch
    -0.67
    POSITIVE LOGITS
    principalColumn
    0.73
     taken
    0.70
    tagHelperRunner
    0.69
    RenderAtEndOf
    0.67
     ActivatedRoute
    0.67
     put
    0.65
     given
    0.64
     brought
    0.62
     properly
    0.61
    nloa
    0.60
    Activations Density 1.025%

    No Known Activations