© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-2B
    4. Residual Stream - 16k
    5. 18-GEMMASCOPE-RES-16K
    6. 7350
    Prev
    Next
    INDEX
    Explanations

    consistency or similarity in systems or structures across different contexts

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

    Something remaining the same or unchanged

    np_acts-logits-general · gemini-2.0-flash

    unchanged or the same

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.18.hook_resid_post
    Hook Layer
    18
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    niająca
    -0.61
    ighborhood
    -0.60
    ftagPool
    -0.59
     Numerade
    -0.54
    Multiple
    -0.51
    aporan
    -0.49
    ksikon
    -0.48
    他にも
    -0.48
    waitKey
    -0.48
    спонд
    -0.47
    POSITIVE LOGITS
     unchanged
    2.08
    Same
    1.62
    same
    1.59
     same
    1.58
     Same
    1.48
     unaltered
    1.46
    SAME
    1.29
     identical
    1.27
    変わらず
    1.27
     dezelfde
    1.26
    Activations Density 0.588%

    No Known Activations