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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-3-12B
    3. 24-GEMMASCOPE-2-RES-16K
    4. 12152
    Prev
    Next
    INDEX
    Explanations

    command usage

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

    The neuron strongly activates on imperative, instruction-style tokens—especially command/tool names and verbs telling the reader what to do.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    technical or instructional content, particularly step-by-step procedures and software/computer-related documentation.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408

    technical computing and networking instructions, especially Windows command-line and configuration terminology within how-to style explanations.

    oai_token-act-pair · gpt-5Triggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_24_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
    납
    0.68
    oblig
    0.64
    ใน
    0.58
     Corporate
    0.58
    Cand
    0.57
    removeFrom
    0.57
     वि
    0.56
    horm
    0.56
    JsonProperty
    0.55
    contract
    0.55
    POSITIVE LOGITS
    命令行
    1.29
     argv
    1.16
     comandos
    1.13
    程序的
    1.04
     프로그램을
    1.03
     programma
    1.02
     程序
    1.00
     executar
    0.99
     programas
    0.98
    コマンド
    0.98
    Activations Density 0.955%

    No Known Activations