© 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. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    5. Llama3.1-8B-IT
    7. Resid Post - 131k
    9. 11-RESID-POST-AA
    11. 38930
    Prev
    Next
    INDEX
    Explanations

    code, programming

    np_max-act · gemini-2.0-flash

    This neuron primarily detects the placeholder token NAME_1.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8

    tokens that are redacted or anonymized placeholders for names/entities (e.g., NAME_x).

    oai_token-act-pair · gpt-5-miniTriggered by @vetterc0
    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
     Cain
    -0.06
     pie
    -0.06
     escort
    -0.06
    ्वय
    -0.06
    Mayor
    -0.06
    ，不过
    -0.06
    Taylor
    -0.06
    �
    -0.05
     getData
    -0.05
     bour
    -0.05
    POSITIVE LOGITS
    άλυ
    0.08
    -analysis
    0.07
    entric
    0.07
    -Americ
    0.07
    -create
    0.07
    -Allow
    0.07
    ρηση
    0.07
    드립니다
    0.06
     στην
    0.06
     некоторые
    0.06
    Activations Density 0.341%

    No Known Activations