© 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. 12921
    Prev
    Next
    INDEX
    Explanations

    building materials, managing conditions, specific terms

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

    The neuron consistently lights up on content‐bearing words—nouns, adjectives, verbs, technical terms, proper names, and numbers—while ignoring most function words.

    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
     Because
    -1.17
     comforting
    -1.06
     Since
    -1.01
    Lá
    -1.00
     rudimentary
    -0.99
     meticulous
    -0.99
     decreases
    -0.97
     isoto
    -0.97
     urso
    -0.96
     disconcer
    -0.96
    POSITIVE LOGITS
    
    1.14
     agarre
    1.06
    lamps
    1.04
     และ
    1.03
     cementerio
    1.02
     différentes
    1.01
     pomys
    1.01
     lauded
    0.99
    护理
    0.99
     agujeros
    0.97
    Activations Density 0.084%

    No Known Activations