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-2-2B
    4. Residual Stream - 65k
    5. 20-GEMMASCOPE-RES-65K
    6. 943
    Prev
    Next
    INDEX
    Explanations

    syntax and structure of programming functions

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

    end brackets and parentheses in code

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

    mail sending logic

    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
    bootstrapcdn
    -0.80
     CreateTagHelper
    -0.65
     оригіналу
    -0.64
    {{/
    -0.63
    rrggbb
    -0.61
    NUMX
    -0.60
    StructEnd
    -0.59
    audiovisuel
    -0.59
     HttpNotFound
    -0.57
     Crim
    -0.57
    POSITIVE LOGITS
    nictví
    0.53
    ymce
    0.49
    verse
    0.49
    feira
    0.47
    jestel
    0.45
    ia
    0.45
     vs
    0.45
     schlecht
    0.44
     tuot
    0.43
    CROSSTALK
    0.42
    Activations Density 0.021%

    No Known Activations