© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-3-27B-IT
    3. 37-GEMMASCOPE-2-TRANSCODER-262K
    4. 204510
    Prev
    Next
    INDEX
    Explanations

    The user wants me to explain the behavior of a neuron based on four lists.The explanation must be concise (3 to 20 words), a phrase, and capture what the neuron detects or predicts by finding patterns.Here's a breakdown of the lists and potential patterns:1. **MAX_ACTIVATING_TOKENS**: * 'an' * ')' * 'Safety' * '.' * '.' * 'Fire' * 'Include' * 'called' * 'and' This list contains punctuation, conjunctions, and words like "Safety" and "Fire".2. **TOKENS_AFTER_MAX_ACTIVATING_TOKEN**: * 'application' * '*' * '**' * 'of' * 'es' * 'existing' * 'Ext' * 'sketches' * 'continuous' * 'maintenance'

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-27b-it/transcoder_all/layer_37_width_262k_l0_small_affine
    Prompts (Dashboard)
    238,145 prompts, 512 tokens each
    Dataset (Dashboard)
    lmsys + oasst1
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     specification
    1.34
     spec
    1.20
    Specification
    1.15
    specification
    1.14
     specifications
    1.13
    Spec
    1.13
     Specification
    1.10
     специ
    1.10
     specifying
    1.07
     espec
    1.05
    POSITIVE LOGITS
     Programme
    0.43
     Deb
    0.40
     deb
    0.40
    鹅
    0.39
     ká
    0.38
     Program
    0.38
     దృ
    0.38
     Programmes
    0.38
    Deb
    0.38
     Dain
    0.38
    Activations Density 0.022%

    No Known Activations