© 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. 2-GEMMASCOPE-2-TRANSCODER-262K
    4. 160761
    Prev
    Next
    INDEX
    Explanations

    environment, government, or social"},"governance"},{"The Neuron Behavior": "environment or social context"},{"The Neuron Behavior": "environment, social, or governance aspects"},{"The Neuron Behavior": "environment, social, or governance contexts"},{"The Neuron Behavior": "environment, social, or governance"},{"The Neuron Behavior": "environment, social, or governance"}{"The Neuron Behavior": "environment or social"},{"The Neuron Behavior": "environment, social/governance"},{"The Neuron Behavior": "environment or social"}There's a clear pattern here. The MAX_ACTIVATING_TOKENS are often parts of words like "environment", "government", or "social". The TOKENS_AFTER_MAX_ACTIVATING_TOKEN strongly support this, showing "nment" (from environment/government), "ion" (from social/governance), "Cafe" (from social context), etc.The phrase should capture this."environment or social""environment, social, governance""environment, social, or governance context"Let's refine based on rules:- Concise (3-20 words).- Specific.- No "tokens", "patterns".- No "This neuron detects/predicts"."environment, social, governance" fits well."environment, social, or governance" also works.Let's look at the example texts.- "enviornment" (typo)- "governance"- "Enviornment"- "enviornmentalism"- "enviornmentalism"The key concepts are "environment", "social", and "governance".The list TOP_POSITIVE_LOGITS has `waterslide`, which could be related to a specific type of environmental activity, but the overall pattern from the other lists is stronger.The most straightforward pattern is the compound concepts."environment, social, governance" is 3 words.environment, social, governance

    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_2_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
    ↵↵
    1.33
    6
    0.97
    7
    0.95
    '
    0.92
    ة
    0.89
    ını
    0.89
    một
    0.88
    ي
    0.87
    𝐒
    0.87
    斯
    0.86
    POSITIVE LOGITS
     a
    1.41
    ä
    1.23
    ができる
    1.09
    ästä
    1.00
    á
    0.96
    c
    0.95
    ik
    0.93
     an
    0.90
    اک
    0.90
     waterslide
    0.89
    Activations Density 0.000%

    No Known Activations