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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-9B-IT
    7. Residual Stream - 16k
    9. 20-GEMMASCOPE-RES-16K
    11. 12120
    Prev
    Next
    INDEX
    Explanations

    phrases that indicate a need for introspection or self-reflection

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

    Follows "too", "is", "but", or "upon"

    np_acts-logits-general · gemini-2.0-flash

    so too upon

    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
     Houſe
    -0.66
     whoſe
    -0.65
    Everybody
    -0.65
    everybody
    -0.63
     houſe
    -0.63
     raiſ
    -0.61
     Jefus
    -0.59
     GUYS
    -0.59
     Majefty
    -0.59
     Everybody
    -0.58
    POSITIVE LOGITS
     său
    0.56
     upon
    0.48
     shall
    0.46
     vermag
    0.46
     iż
    0.45
    这般
    0.44
     sought
    0.44
     tevens
    0.43
     presso
    0.43
     Upon
    0.42
    Activations Density 1.485%

    No Known Activations