© 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
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B-IT
    4. Residual Stream - 131k
    5. 9-GEMMASCOPE-RES-131K
    6. 107448
    Prev
    Next
    INDEX
    Explanations

    phrases related to suggestions or actions for improvement

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

    how to fix or improve something

    np_acts-logits-general · gemini-2.5-flash-lite

    It detects instructional or actionable content—phrases that introduce steps, tips, how-to guidance, or things you can do.

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 9-gemmascope-res-131k
    Configuration
    google/gemma-scope-9b-it-res/layer_9/width_131k/average_l0_121
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.9.hook_resid_post
    Hook Layer
    9
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     MainAxisSize
    -0.45
     Divulgação
    -0.45
    ValueStyle
    -0.39
     hábiles
    -0.37
     amarillas
    -0.37
    encodeWith
    -0.37
     Kelurahan
    -0.37
    控制器
    -0.37
     maíz
    -0.36
     confé
    -0.35
    POSITIVE LOGITS
     Trick
    0.60
    rungsseite
    0.55
    Trick
    0.53
     Tricks
    0.53
    RTGC
    0.52
    ANTLR
    0.50
    Heck
    0.47
    řes
    0.47
    VIAF
    0.47
    tricks
    0.47
    Activations Density 0.075%

    No Known Activations