© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    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. 203220
    Prev
    Next
    INDEX
    Explanations

    This neuron seems to be recognizing concepts related to lightweight materials and their application, particularly in the automotive industry, often contrasted with heavier or traditional materials like steel.Let's try to find a concise phrase.- **MAX_ACTIVATING_TOKENS**: lighter, lightweight, automotive, strong, strength, of- **TOKENS_AFTER_MAX_ACTIVATING_TOKEN**: materials, carbon, vehicles, structure, steelThe combination of "light" (lighter, lightweight) and "materials" (materials, carbon, steel, automotive) is strong. The `TOKENS_AFTER_MAX_ACTIVATING_TOKEN` list shows words like "materials", "carbon", "vehicles", "structure", and "steel".lightweight automotive materials

    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
     लीक
    0.37
    Italia
    0.36
    ctor
    0.36
    λαν
    0.36
     gỗ
    0.34
    ewood
    0.34
    깃
    0.34
    轱
    0.34
    cstr
    0.34
    ンテ
    0.34
    POSITIVE LOGITS
    ベイ
    0.43
    Reduction
    0.41
     جنہوں
    0.39
     menurunkan
    0.38
     pengurangan
    0.38
    भव
    0.38
     Reduction
    0.37
     பெயரை
    0.37
    yli
    0.37
     Macaulay
    0.37
    Activations Density 0.001%

    No Known Activations