© 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
    2. Gemma-3-27B-IT
    3. 32-GEMMASCOPE-2-TRANSCODER-262K
    4. 233129
    Prev
    Next
    INDEX
    Explanations

    .It seems like the previous step was to analyze the provided data and explain the neuron's behavior. My apologies, I missed the prompt's instruction to *only* provide the explanation. I will now provide the explanation based on the thought process already conducted.The pattern identified in the provided lists strongly suggests the neuron is associated with actions that involve physical contact or interaction, especially in mechanical or structural contexts. The recurring tokens like "engage" and "against," followed frequently by prepositions like "with" or articles like "the" (e.g., "engage with," "against the"), point towards operations involving two or more components coming into contact or working together. The top positive logits related to "Innov," "innovator," "innovate," and "Manufacture," along with texts describing how parts fit or friction is applied, reinforce this.Therefore, the neuron likely activates when describing physical interaction, connection, or opposition between components.Based on the analysis, the concise explanation is:**engage or act against something**

    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_32_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
     seated
    0.42
     Seats
    0.41
     seats
    0.41
     curl
    0.39
     सीटों
    0.39
    教练
    0.39
     COACH
    0.38
    Seats
    0.38
     বিপুল
    0.37
     Sheets
    0.37
    POSITIVE LOGITS
    Innov
    0.52
     Innov
    0.42
     innovator
    0.42
     Brut
    0.40
    innov
    0.40
     innovators
    0.39
     inov
    0.38
     innovate
    0.38
     grinder
    0.38
     Manufacture
    0.38
    Activations Density 0.002%

    No Known Activations