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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & 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. 3538
    Prev
    Next
    INDEX
    Explanations

    expressions of admission, confession, or acknowledgment of wrongdoing or sensitive subjects

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

    Admitting or conceding something

    np_acts-logits-general · gemini-2.0-flash

    admit or concede

    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
    esgue
    -0.65
     viewDidLoad
    -0.60
     wherefore
    -0.57
    vábbi
    -0.57
     typelib
    -0.57
    newArrayList
    -0.57
    voegd
    -0.56
    BuildContext
    -0.55
    RSpec
    -0.55
    UrlResolution
    -0.55
    POSITIVE LOGITS
     admit
    2.78
     admitting
    2.59
     admitted
    2.54
     admission
    2.47
     admits
    2.41
     Admit
    2.25
     Admission
    2.07
     acknowledge
    2.04
     admissions
    1.99
    admission
    1.98
    Activations Density 0.412%

    No Known Activations