© 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. 84183
    Prev
    Next
    INDEX
    Explanations

    I thought / thinks

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

    The neuron is keyed to dummy‐subject “it” constructions (e.g. “it might be…,” “I thought it would…”), i.e. the pronoun “it” when used as a filler subject in speculative or introductory clauses.

    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
    Their
    -1.04
    skrä
    -1.00
     februari
    -0.98
    jährige
    -0.97
    rô
    -0.96
    丨
    -0.95
    akaan
    -0.94
    ☟
    -0.94
    ੴ
    -0.93
     не
    -0.92
    POSITIVE LOGITS
     ought
    1.40
     might
    1.16
     apt
    1.13
     better
    1.13
     about
    1.11
     too
    1.10
     appropriate
    1.09
     geeigne
    1.08
     maybe
    1.07
     best
    1.05
    Activations Density 0.046%

    No Known Activations