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. 4662
    Prev
    Next
    INDEX
    Explanations

    references to time durations in the context of events

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

    "Few" followed by a unit of time or distance

    np_acts-logits-general · gemini-2.0-flash

    few and time/quantity

    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
     algum
    -0.55
     alguna
    -0.53
     suatu
    -0.53
     algún
    -0.50
     alguma
    -0.49
    stoffe
    -0.48
    INDIAN
    -0.47
     helft
    -0.46
    EndContext
    -0.45
     somehow
    -0.45
    POSITIVE LOGITS
     hundred
    1.34
     dozen
    1.28
     thousand
    1.10
     weeks
    1.08
    hundred
    1.03
    dozen
    1.02
     hundert
    0.92
    thousand
    0.92
     wenige
    0.88
     months
    0.87
    Activations Density 0.121%

    No Known Activations