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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B-IT
    4. Residual Stream - 131k
    5. 20-GEMMASCOPE-RES-131K
    6. 5743
    Prev
    Next
    INDEX
    Explanations

    negative statements or phrases expressed in the form of "shouldn't," "can't," or "don't."

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

    should not or shouldn't

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k
    Configuration
    google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    131,072
    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
    fara
    -0.55
    }$​
    -0.49
    endcsname
    -0.48
    AFX
    -0.48
    aré
    -0.47
     |
    
    -0.46
    \"");
    -0.45
    ."</
    -0.44
    unknownFields
    -0.44
    )}</
    -0.44
    POSITIVE LOGITS
    should
    1.04
    Should
    1.01
     should
    1.01
     Should
    1.01
     SHOULD
    0.93
    hould
    0.77
    不应该
    0.67
    ShouldBe
    0.65
     ought
    0.61
     bör
    0.60
    Activations Density 0.010%

    No Known Activations