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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    5. Llama3.1-8B-IT (Instruct)
    7. Resid Post - 131k
    9. 11-RESID-POST-AA
    11. 85802
    Prev
    Next
    INDEX
    Explanations

    math equations

    np_max-act · gemini-2.0-flash

    The neuron detects tokens that are part of mathematical expressions—particularly function names (like sin, cos, e) and exponent notation in calculus contexts.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1
    Dataset (Dashboard)
    Various
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.11.hook_resid_post
    Architecture
    standard
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    рабат
    -0.06
    (hex
    -0.06
    ็ง
    -0.06
     применения
    -0.06
     Weber
    -0.06
     Psychiatry
    -0.06
    ause
    -0.06
    /fire
    -0.06
     开始
    -0.06
     für
    -0.06
    POSITIVE LOGITS
     François
    0.07
    重複重複
    0.07
     mailing
    0.06
     Spencer
    0.06
    newInstance
    0.06
    Civil
    0.06
    。
    0.06
     Adjustment
    0.06
     Performs
    0.06
    ента
    0.06
    Activations Density 0.008%

    No Known Activations