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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlogSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B-IT
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 8394
    Prev
    Next
    INDEX
    Explanations

    instances of the pronoun "I" and related personal statements

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

    Repeated use of first-person pronouns ("I" or "I'm") at the beginning of text passages, often introducing a personal statement or narrative.

    eleuther_acts_top20 · claude-3-5-haiku-20241022Triggered by @iarcuschin

    first-person pronouns and helping verbs (particularly "I" and auxiliary verbs like "would", "was", "need", "can").

    oai_token-act-pair · claude-3-5-haiku-20241022Triggered by @iarcuschin
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91
    Prompts (Dashboard)
    24,576 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
     ویکی‌پدی
    -0.69
    windowFixed
    -0.62
    parsedMessage
    -0.61
    FDRE
    -0.55
    AccessorTable
    -0.54
     jspb
    -0.54
    verifyException
    -0.54
     Roskov
    -0.54
     Италијани
    -0.52
    Rüyada
    -0.52
    POSITIVE LOGITS
     queſto
    0.50
    rungsseite
    0.49
     imagui
    0.48
    уда
    0.48
    ﹧
    0.47
     好文分享
    0.47
     zwiſchen
    0.47
    Чыгана
    0.46
    mathrm
    0.46
     vooz
    0.45
    Activations Density 0.195%

    No Known Activations