© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B
    4. Residual Stream - 16k
    5. 28-GEMMASCOPE-RES-16K
    6. 11227
    Prev
    Next
    INDEX
    Explanations

    functionality and user actions related to software features and options

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

    Follows "can" or "to" and a verb

    np_acts-logits-general · gemini-2.0-flash

    can do actions

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B @ 28-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-pt-res/layer_28/width_16k/average_l0_119
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.28.hook_resid_post
    Hook Layer
    28
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    帖最后由
    -0.43
    ngOn
    -0.41
    __;
    -0.40
     OMITTED
    -0.39
    twimg
    -0.39
    ChildScrollView
    -0.38
    úrate
    -0.37
    fant
    -0.36
     Bomb
    -0.36
     wrapper
    -0.35
    POSITIVE LOGITS
     customize
    0.63
    directly
    0.59
     interchange
    0.54
     access
    0.53
     interact
    0.52
     directly
    0.52
     customise
    0.52
    UpInside
    0.52
     manipulate
    0.52
     easily
    0.51
    Activations Density 0.471%

    No Known Activations