© 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
    2. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    3. Llama3.1-8B-IT (Instruct)
    4. Resid Post - 131k
    5. 11-RESID-POST-AA
    6. 30448
    Prev
    Next
    INDEX
    Explanations

    conversations

    np_max-act · gemini-2.0-flash

    The neuron is mainly activating on speaker‐label tokens (e.g. “NAME_1,” “NAME_2,” etc.), i.e. it flags the utterance tags identifying who’s speaking rather than the content words.

    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
    <!
    -0.06
    ',↵↵
    -0.06
    ,st
    -0.06
    現在
    -0.06
    ниці
    -0.06
    ,现在
    -0.06
     ginger
    -0.06
    pus
    -0.06
    .getAll
    -0.06
     ev
    -0.06
    POSITIVE LOGITS
    كات
    0.06
     ألمان
    0.06
     logistical
    0.06
     cooperating
    0.06
     Dive
    0.06
    jez
    0.06
    _FINE
    0.06
     consenting
    0.06
    يدة
    0.06
    okoj
    0.06
    Activations Density 0.019%

    No Known Activations