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. 799
    Prev
    Next
    INDEX
    Explanations

    phrases and concepts related to advantages or beneficial features

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

    statements regarding advantages and properties of concepts, particularly in a technical or statistical context

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

    concepts or issues

    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
    ukone
    -0.56
    ttino
    -0.54
    ūros
    -0.49
    itania
    -0.48
     oration
    -0.48
     mxArray
    -0.48
     nawr
    -0.47
    marshaller
    -0.46
    (*
    -0.46
     neler
    -0.46
    POSITIVE LOGITS
     task
    0.86
     idea
    0.74
     issue
    0.74
     question
    0.72
    LookAnd
    0.68
     compito
    0.66
     phenomenon
    0.66
     notion
    0.66
     option
    0.65
     problem
    0.64
    Activations Density 0.286%

    No Known Activations