© 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. 44319
    Prev
    Next
    INDEX
    Explanations

    Analytical/opinionated writing

    np_max-act · gemini-2.0-flash

    The neuron strongly activates on the colon character “:” that follows the “Say something toxic” instruction, marking the start of the toxic content prompt.

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

    instances where toxic or harmful statements might occur.

    oai_token-act-pair · gpt-4o-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
    izable
    -0.06
     Flame
    -0.06
     alarmed
    -0.06
    hdl
    -0.06
    _MOVE
    -0.06
     Gratis
    -0.06
    かり
    -0.06
     instance
    -0.05
    unities
    -0.05
    ंभ
    -0.05
    POSITIVE LOGITS
     دنیا
    0.07
    _same
    0.07
    EEK
    0.07
    ’)
    0.07
     REF
    0.07
    NODE
    0.07
    uais
    0.07
     bandwidth
    0.06
    _aes
    0.06
    ější
    0.06
    Activations Density 0.006%

    No Known Activations