Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B-IT
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 11029
    Prev
    Next
    INDEX
    Explanations

    phrases associated with quality and performance improvements

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

    Adjectives after "better", "lack", "high", or "good"

    np_acts-logits-general · gemini-2.0-flash

    adjective followed by noun

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91
    Prompts (Dashboard)
    24,576 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
     and
    -0.56
     either
    -0.42
     also
    -0.40
     Landschafts
    -0.39
     or
    -0.38
     Töchter
    -0.36
    Patri
    -0.36
    isinstance
    -0.36
     however
    -0.35
    either
    -0.35
    POSITIVE LOGITS
    +#+#
    0.75
    ьаж
    0.69
     חיצוניים
    0.67
     BoxFit
    0.60
    aarrggbb
    0.60
     estekak
    0.59
    脚注の使い方
    0.58
    IntoConstraints
    0.56
     وتسجيلات
    0.56
    ProcessEvent
    0.56
    Activations Density 0.815%

    No Known Activations