© 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. 10-GEMMASCOPE-RES-131K
    7. 112929
    Prev
    Next
    INDEX
    Explanations

    THE NEXT AGE

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

    The neuron fires on isolated, out-of-context tokens that aren’t ordinary English words—e.g. foreign-language words (like Latvian “tika”) or code‐markup fragments (like “/>”).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
     doesnt
    -2.63
    it
    -2.42
    呖
    -2.25
    tividade
    -2.20
    er
    -2.16
     尔
    -2.14
    羋
    -2.13
    ቍ
    -2.13
     ihti
    -2.11
     afir
    -2.11
    POSITIVE LOGITS
     tetapi
    2.72
    .’
    2.39
            
    2.34
    '
    2.31
     Previously
    2.28
    1
    2.20
     serán
    2.17
                        
    2.11
     individuellen
    2.09
     лишь
    2.08
    Activations Density 0.000%

    No Known Activations