© 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. Qwen3-1.7B
    5. 26-LLAMASCOPE-2-LORSA-16K-K64
    7. 1078
    Prev
    Next
    INDEX
    Explanations

    Method 5 – the TOP_NEGATIVE_LOGITS are all variations of “tool”. The neuron strongly suppresses any “tool” token. **do not say tools**

    unknown · unknown
    New Auto-Interp
    Top Features by Cosine Similarity
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    工具
    -70.00
     tools
    -64.50
     tool
    -62.75
     Tools
    -57.75
     Tool
    -54.75
     TOOL
    -51.25
    Tools
    -51.00
    tools
    -50.25
    Tool
    -50.00
    -tools
    -49.25
    POSITIVE LOGITS
    phy
    14.81
    痣
    14.63
    matrix
    14.50
    sin
    14.13
    .sin
    14.13
    耶稣
    13.69
    彘
    13.56
    Jesus
    13.06
    罪
    13.06
    ọi
    12.88
    Activations Density 0.130%

    No Known Activations