© 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
    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. 9170
    Prev
    Next
    INDEX
    Explanations

    answers; responses

    np_max-act · gemini-2.0-flash

    This neuron detects meta-instructions or role-play directives—phrasing where the user tells the model how to answer (e.g. “Answer as … in the form of …”).

    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
     letto
    -0.07
     Deadpool
    -0.07
    ade
    -0.06
    識
    -0.06
     cov
    -0.06
     computations
    -0.06
     gypsum
    -0.06
    nego
    -0.06
     Mariners
    -0.06
    _symbol
    -0.06
    POSITIVE LOGITS
     -->
↵
    0.06
    "<?
    0.06
    (flag
    0.06
     :::
    0.06
    .Border
    0.06
    ;!
    0.06
     kaliteli
    0.06
    	sound
    0.06
    ディ
    0.06
     glob
    0.06
    Activations Density 0.019%

    No Known Activations