© 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 EvalsExportsinterp-engineNEWAPI 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. 57629
    Prev
    Next
    INDEX
    Explanations

    believe

    np_max-act · gemini-2.0-flash

    The neuron fires on evaluative words and phrases that convey conviction or authenticity—i.e. reviewer language like “believe,” “authenticity,” “confidence,” “fine job,” “brings a feeling of…,” etc.

    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
     lighter
    -0.07
     zásad
    -0.07
    کز
    -0.07
    everything
    -0.06
     edilmesi
    -0.06
    kontakte
    -0.06
    CAP
    -0.06
     doctr
    -0.06
    ськ
    -0.06
    _with
    -0.06
    POSITIVE LOGITS
    720
    0.07
     defy
    0.06
    .faces
    0.06
    ivals
    0.06
    游戏
    0.06
     食
    0.06
    폰
    0.06
     господар
    0.06
     천
    0.06
     separated
    0.06
    Activations Density 0.018%

    No Known Activations