© 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
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 75104
    Prev
    Next
    INDEX
    Explanations

    suffix/ending after learned roots

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

    The neuron consistently fires on short subword fragments that begin complex, domain‐specific or technical terms (i.e. on the prefixes of long specialized 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
    になりますが
    -1.32
     transferir
    -1.31
     erhi
    -1.30
    malıdır
    -1.28
    madı
    -1.25
    Hoi
    -1.24
    grats
    -1.24
    makeat
    -1.23
    inspir
    -1.21
    Crítica
    -1.21
    POSITIVE LOGITS
    V
    1.16
    】：
    1.16
    T
    1.16
    P
    1.16
     '
    1.13
    e
    1.10
    W
    1.10
    ▿
    1.09
    R
    1.09
    ed
    1.09
    Activations Density 0.194%

    No Known Activations