© 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. 5955
    Prev
    Next
    INDEX
    Explanations

    delayed information

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

    The neuron is generally flagging unusually rare or specialized content words (e.g. “praise,” “types,” “Manufacturer,” “to” in the Quranic phrase “praise be to”) – essentially spotting low-frequency, domain-specific tokens.

    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_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
    ྜ
    0.81
     isEmpty
    0.76
    𝑏
    0.76
     OnInit
    0.75
     WENO
    0.74
     uprisings
    0.73
     condiciones
    0.72
     войны
    0.72
     смерти
    0.72
    𝑐
    0.72
    POSITIVE LOGITS
    ENT
    0.82
    EST
    0.81
    EM
    0.79
    erweise
    0.79
    PH
    0.78
     इतर
    0.76
    Pa
    0.75
    ERR
    0.75
    eV
    0.75
     etwa
    0.75
    Activations Density 0.000%

    No Known Activations