© 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. 54247
    Prev
    Next
    INDEX
    Explanations

    -ise/-ize suffixes

    np_max-act · gemini-2.0-flash

    This neuron lights up on words expressing transformative change—most strongly on “revolutionize” (and its “-ize” suffix) and, to a lesser extent, on related change verbs like “changes” and phrases indicating a new “way.”

    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
     šť
    -0.06
    眉
    -0.06
     expresses
    -0.06
    �
    -0.06
     adm
    -0.06
    mony
    -0.06
     brightest
    -0.06
     TWO
    -0.06
     older
    -0.05
     kü
    -0.05
    POSITIVE LOGITS
    _method
    0.07
     Changed
    0.07
     revolutionary
    0.07
    时
    0.07
    设计器
    0.07
    _missing
    0.07
     Favorites
    0.07
     Overlay
    0.07
    _components
    0.07
     cnt
    0.07
    Activations Density 0.022%

    No Known Activations