© 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 - 65k
    5. 20-GEMMASCOPE-RES-65K
    6. 1546
    Prev
    Next
    INDEX
    Explanations

    non-alphanumeric characters and punctuation marks

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

    punctuation and formatting symbols, particularly in the context of social media or online communication

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

    Please, thank you, review

    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-65k
    Configuration
    google/gemma-scope-2b-pt-res/layer_20/width_65k/average_l0_114
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    65,536
    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
    GenerationType
    -0.61
    ➟
    -0.58
    NOPQRST
    -0.57
    脚注の使い方
    -0.55
    Newswire
    -0.55
    省市镇
    -0.54
    zzleHttp
    -0.54
    нциклопедия
    -0.51
    errHandler
    -0.51
    indd
    -0.50
    POSITIVE LOGITS
     Enjoy
    0.91
     Please
    0.85
     credits
    0.84
     please
    0.81
     PLEASE
    0.79
    Enjoy
    0.78
    Please
    0.78
     ENJOY
    0.78
     Credits
    0.76
     NOTE
    0.76
    Activations Density 0.052%

    No Known Activations