© 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-2-9B-IT
    4. Residual Stream - 16k
    5. 20-GEMMASCOPE-RES-16K
    6. 6701
    Prev
    Next
    INDEX
    Explanations

    proper nouns, particularly names of people and places

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

    followed by names

    np_acts-logits-general · gemini-2.0-flash

    surnames ending in son, ckus, gers, uck

    np_acts-logits-general · gemini-2.5-flash-lite
    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
     just
    -0.46
     not
    -0.38
     very
    -0.36
     indeed
    -0.36
     really
    -0.36
     already
    -0.36
     still
    -0.36
     rather
    -0.35
     ever
    -0.35
     well
    -0.35
    POSITIVE LOGITS
    windowFixed
    0.71
     surla
    0.70
    iſen
    0.66
     Numerade
    0.65
     pinulongan
    0.65
    pecabe
    0.64
    nezeu
    0.63
    ésultats
    0.61
    ſammen
    0.59
    RTEX
    0.59
    Activations Density 0.788%

    No Known Activations