© 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
    3. Gemma-3-27B-IT
    5. 40-GEMMASCOPE-2-TRANSCODER-262K
    7. 28961
    Prev
    Next
    INDEX
    Explanations

    The user wants a concise explanation (3-20 words) for a neuron's behavior.I need to analyze the provided lists: `TOP_POSITIVE_LOGITS`, `TOP_ACTIVATING_TEXTS`.The lists `MAX_ACTIVATING_TOKENS` and `TOKENS_AFTER_MAX_ACTIVATING_TOKEN` were empty in the input, so I will focus on the logits and texts.`TOP_POSITIVE_LOGITS`:- compensation- condens- buildup- ့် (likely a character or part of a word)- प्रया (likely characters from a language like Hindi)- atribut (likely 'attribute')- آیات (likely Arabic for 'verses' or 'signs')- Acknowledgement- condensate- 築 (likely Chinese character)`TOP_ACTIVATING_TEXTS`:The texts are snippets from code files and documentation, often related to programming, API development, and frameworks.`csharp` code snippets

    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_40_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
     thirteenth
    0.43
     eleventh
    0.38
    Import
    0.37
    berkeley
    0.37
    eighth
    0.37
     fourteenth
    0.37
    Parser
    0.36
     Tenth
    0.36
    亮度
    0.35
     વી
    0.35
    POSITIVE LOGITS
     compensation
    0.41
     condens
    0.39
     buildup
    0.38
    ့်
    0.38
     प्रया
    0.37
     atribut
    0.37
     آیات
    0.37
    Acknowledgement
    0.36
     condensate
    0.36
    築
    0.35
    Activations Density 0.000%

    No Known Activations