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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B-IT
    4. Residual Stream - 131k
    5. 20-GEMMASCOPE-RES-131K
    6. 108434
    Prev
    Next
    INDEX
    Explanations

    instances of the word "somehow" to indicate uncertainty or mystery

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

    unexpected connections or outcomes

    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-131k
    Configuration
    google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    131,072
    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
    PreExecute
    -0.55
     Reserve
    -0.43
     reserve
    -0.42
    Reserve
    -0.41
     praco
    -0.40
     carteira
    -0.39
     list
    -0.38
     List
    -0.38
     HasFactory
    -0.38
     kill
    -0.36
    POSITIVE LOGITS
    Somehow
    0.95
     Somehow
    0.92
     somehow
    0.90
    WriteTagHelper
    0.76
    xically
    0.68
     magically
    0.63
     miraculously
    0.61
    Somewhere
    0.58
    Someday
    0.57
    addPreferredGap
    0.57
    Activations Density 0.006%

    No Known Activations