© 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
    3. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    5. Llama3.1-8B-IT (Instruct)
    7. Resid Post - 131k
    9. 11-RESID-POST-AA
    11. 85720
    Prev
    Next
    INDEX
    Explanations

    Sentence beginnings

    np_max-act · gemini-2.0-flash

    This neuron activates on the standard first-person “I’m sorry, but as an AI language model…” style apology/refusal phrases.

    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
    いう
    -0.07
    :[
    -0.06
    errMsg
    -0.06
     root
    -0.06
    、と
    -0.06
    熟
    -0.06
    Sun
    -0.06
    跟
    -0.06
    ุธ
    -0.06
     ⟨
    -0.06
    POSITIVE LOGITS
    ERSHEY
    0.07
    PAR
    0.06
    BA
    0.06
     Petite
    0.06
    onna
    0.06
     regulators
    0.06
    ندگی
    0.06
    hurst
    0.06
    avadoc
    0.06
     EntityType
    0.06
    Activations Density 0.016%

    No Known Activations