© 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-2-2B
    4. Residual Stream - 16k
    5. 22-GEMMASCOPE-RES-16K
    6. 3192
    Prev
    Next
    INDEX
    Explanations

    sentences discussing opinions or critiques about a football manager's decisions and team performance

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

    defending, defense, or pressing

    np_acts-logits-general · gemini-2.0-flash

    defensive football actions

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 22-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_22/width_16k/average_l0_72
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.22.hook_resid_post
    Hook Layer
    22
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     morada
    -0.49
    ̈́
    -0.49
    amoan
    -0.48
     mulus
    -0.47
     IBOutlet
    -0.45
    Décès
    -0.45
    DebuggerNonUser
    -0.44
    usalem
    -0.44
     cref
    -0.44
     osu
    -0.44
    POSITIVE LOGITS
     attacking
    0.64
     compactness
    0.63
     pressing
    0.63
     compact
    0.61
     attack
    0.59
     Pressing
    0.58
     possession
    0.57
    pressing
    0.57
     attacks
    0.57
     Attack
    0.55
    Activations Density 0.080%

    No Known Activations