© 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. 31-GEMMASCOPE-2-TRANSCODER-262K
    4. 77435
    Prev
    Next
    INDEX
    Explanations

    This neuron seems to be about identifying used car descriptions. It looks for common phrases and details found in ads for selling vehicles, especially concerning their condition and specifications.Specifically, I noticed several key elements:1. **Condition Descriptors:** Phrases like "Good condition," "Runs great," and mentions of "wear and tear" or "no major mechanical issues" appear frequently.2. **Vehicle Specifics/Identifiers:** Tokens like "title," "mileage," and potentially parts of car model names (though not explicitly listed here, they are implied by the context of car ads).3. **Action/Purpose:** Tokens like "gets you from A to B" suggest the car's basic functionality.Given these observations, the neuron appears to be detecting the core components found in classified ads for used cars.**Explanation:**used car sale details

    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_31_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
     Verk
    0.53
     cars
    0.44
     Име
    0.41
     Paddle
    0.41
     конструк
    0.40
    Cars
    0.40
     автомоби
    0.40
     име
    0.39
     automotive
    0.39
    자동차
    0.39
    POSITIVE LOGITS
     라이
    0.37
     puissent
    0.35
    राइड
    0.35
     بكل
    0.35
    colonial
    0.34
    oxys
    0.33
     acidic
    0.33
    rangea
    0.33
    امن
    0.33
     идеально
    0.33
    Activations Density 0.001%

    No Known Activations