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 - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 6645
    Prev
    Next
    INDEX
    Explanations

    occurrences of the letter 'a' in various forms

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

    Following the word "a" in Spanish and French

    np_acts-logits-general · gemini-2.0-flash

    a followed by Spanish/French articles/prepositions

    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-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    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
     itſelf
    -0.98
     myſelf
    -0.92
    theless
    -0.84
     ་་
    -0.83
    IsMutable
    -0.82
     Efq
    -0.82
     Jefus
    -0.82
     Shakspeare
    -0.80
     Eſ
    -0.79
    AndEndTag
    -0.78
    POSITIVE LOGITS
     least
    0.52
     court
    0.51
    parent
    0.49
     fine
    0.48
     partir
    0.48
     level
    0.47
     work
    0.45
     times
    0.45
     scale
    0.45
     
    0.44
    Activations Density 0.036%

    No Known Activations