© 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. 54306
    Prev
    Next
    INDEX
    Explanations

    questions

    np_max-act · gemini-2.0-flash

    The neuron activates on question words and auxiliary/modal verbs (e.g., “Will,” “have,” “What,” “could,” “do,” “should”) that signal a question.

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

    questions seeking causes, potential consequences, or fixes (diagnosis/advice requests with modals and question phrasing)

    oai_token-act-pair · gpt-5Triggered 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
     preferable
    -0.07
    ط
    -0.07
    fg
    -0.07
    برای
    -0.07
    _registers
    -0.06
    worker
    -0.06
    VectorXd
    -0.06
    lu
    -0.06
    ोक
    -0.06
    pir
    -0.06
    POSITIVE LOGITS
    .Mon
    0.06
     �
    0.06
     returnValue
    0.06
    uento
    0.06
     semaine
    0.06
    [source
    0.06
     Archived
    0.06
     Pols
    0.06
     giờ
    0.06
    ekyll
    0.06
    Activations Density 0.090%

    No Known Activations