© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 26541
    Prev
    Next
    INDEX
    Explanations

    license text or programming comments

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

    The neuron fires on numeric or alphanumeric technical identifiers—things like version numbers, standards or method acronyms that include digits (e.g. “MAPK,” “VPS,” “3-Clause,” “Version 2.0,” etc.).

    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
     […]
    -1.15
     them
    -1.09
     mostly
    -1.06
     jsou
    -1.04
     …
    -1.01
    が
    -1.01
    だが
    -1.00
     have
    -1.00
    vereignty
    -0.95
     various
    -0.93
    POSITIVE LOGITS
     alfab
    1.02
     noten
    1.02
     abonnement
    1.02
     guste
    1.00
    んだよね
    0.98
    DOES
    0.98
     aquarelle
    0.96
    Ⅷ
    0.96
     قه
    0.94
     WAVE
    0.93
    Activations Density 0.437%

    No Known Activations