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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & 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. 15848
    Prev
    Next
    INDEX
    Explanations

    punctuation marks and formatting symbols

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

    This neuron appears to activate strongly for sentence-ending punctuation, particularly periods at the ends of sentences. It shows high activation values for periods that conclude complete thoughts or statements.

    oai_token-act-pair · claude-3-5-sonnet-20240620Triggered by @bot

    Text following punctuation

    np_acts-logits-general · gemini-2.0-flash

    personal opinions and conversational starters

    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
    GEBURTSDATUM
    -1.14
     ―――――
    -1.11
     ſind
    -1.03
     NDEBUG
    -1.02
     ſche
    -1.01
     Majefty
    -1.01
     iſt
    -0.99
    Билгалдахарш
    -0.98
     uſ
    -0.95
     neceffary
    -0.94
    POSITIVE LOGITS
     alot
    0.99
     I
    0.90
     thats
    0.88
    Lets
    0.85
     dont
    0.81
     ect
    0.79
     میتوان
    0.74
     we
    0.73
     dimana
    0.73
     you
    0.72
    Activations Density 0.374%

    No Known Activations