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. 64788
    Prev
    Next
    INDEX
    Explanations

    adverbs and adjectives related to movement directions, especially those indicating upward and downward motion

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

    direction and movement

    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
    näm
    -0.48
     geladeira
    -0.43
    assertIs
    -0.43
    IBLIO
    -0.43
    Mix
    -0.42
    StrictEqual
    -0.42
    ergies
    -0.41
     Constitu
    -0.41
     présidenti
    -0.40
    Didn
    -0.40
    POSITIVE LOGITS
     upward
    0.93
     downward
    0.88
    ward
    0.88
     inward
    0.88
     outward
    0.85
     forward
    0.82
     backward
    0.82
     downwards
    0.79
     upwards
    0.79
     towards
    0.78
    Activations Density 0.012%

    No Known Activations