© 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. 12310
    Prev
    Next
    INDEX
    Explanations

    config

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

    The neuron tends to fire strongly on uncommon, irregular or otherwise low‐frequency content words (e.g. unusual verbs, nouns, or adjectives), regardless of their grammatical role.

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

    morphologically marked content words—especially those with common English suffixes like past tense, gerund/participle, plural/3rd-person, and formal derivational endings—indicative of more formal or descriptive language.

    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
    𝐀
    0.80
    ів
    0.79
    <bos>
    0.76
    据
    0.75
    𝐄
    0.75
    𝗦
    0.74
    𝗗
    0.73
    𝗛
    0.73
    رد
    0.72
    𝐈
    0.72
    POSITIVE LOGITS
     Universidade
    0.77
    ش
    0.76
     inflows
    0.74
     cyclopent
    0.74
     baryons
    0.73
     intrusions
    0.73
     homologs
    0.73
    mittance
    0.72
     inhomogeneities
    0.72
    sulf
    0.71
    Activations Density 0.001%

    No Known Activations