© 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-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 79931
    Prev
    Next
    INDEX
    Explanations

    descriptions of subtle states

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

    The neuron detects intense, sentiment-laden descriptive words—especially strong adjectives and adverbs (e.g. “downright,” “deadly,” “predictably,” “numbingly”) used to convey critique or emphasis.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    actéristiques
    -1.49
     différents
    -1.45
    thodoxy
    -1.30
    がとても
    -1.29
    暗暗
    -1.28
    caping
    -1.27
    瞟
    -1.27
     ۳۰
    -1.26
    などは
    -1.24
    reonine
    -1.24
    POSITIVE LOGITS
     seguras
    1.39
     mismas
    1.30
     utterly
    1.20
     both
    1.20
     variadas
    1.13
     profoundly
    1.12
     only
    1.11
     nadru
    1.09
     temperaturas
    1.08
     Masalah
    1.07
    Activations Density 0.029%

    No Known Activations