© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-3-12B
    3. 31-GEMMASCOPE-2-RES-16K
    4. 7237
    Prev
    Next
    INDEX
    Explanations

    achieve goals or results

    np_acts-logits-general · gemini-2.5-flash-lite

    The neuron strongly responds to words and word‐parts that denote achieving or accomplishing goals (e.g., “achieve,” “accomplish,” “achieving,” “results,” “toward”).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_31_width_16k_l0_medium
    Prompts (Dashboard)
    392,802 prompts, 256 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     правила
    0.65
    SearchCV
    0.62
     вопросам
    0.62
     Hurry
    0.59
    последствии
    0.58
    employ
    0.57
     Anschließend
    0.57
    ')])
    0.57
     hurry
    0.56
     вопросы
    0.56
    POSITIVE LOGITS
     achieving
    1.54
     achieves
    1.51
     achieved
    1.49
     achieve
    1.45
     achievement
    1.32
    Goals
    1.26
     accomplishing
    1.26
    達成
    1.26
     accomplishment
    1.25
     Achieving
    1.23
    Activations Density 0.273%

    No Known Activations