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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-9B
    7. Residual Stream - 16k
    9. 25-GEMMASCOPE-RES-16K
    11. 15398
    Prev
    Next
    INDEX
    Explanations

    phrases that emphasize excessive behaviors or conditions

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B @ 25-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-pt-res/layer_25/width_16k/average_l0_114
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.25.hook_resid_post
    Hook Layer
    25
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    OGND
    -0.35
     gesch
    -0.35
     ig
    -0.33
    结
    -0.33
     trat
    -0.32
     kicked
    -0.31
    кновен
    -0.31
    凉
    -0.31
     Mud
    -0.31
     staggered
    -0.30
    POSITIVE LOGITS
     excessive
    1.31
    Excessive
    1.25
     Excessive
    1.20
     exces
    1.18
     excessively
    1.18
     excesso
    1.13
     exceso
    1.12
     berlebihan
    1.08
    过度
    1.00
     excess
    1.00
    Activations Density 0.638%

    No Known Activations