© 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. 6569
    Prev
    Next
    INDEX
    Explanations

    "in" and phrasal verbs

    np_max-act · gemini-2.0-flash

    This neuron detects Arabic root-and-pattern letter sequences—particularly common Arabic word fragments like plural/feminine noun endings (e.g. “–ات” or “–ية”).

    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
     italiani
    -0.07
    Space
    -0.07
    ьому
    -0.07
     por
    -0.07
     většinou
    -0.07
     серьез
    -0.06
    _consumer
    -0.06
     piyas
    -0.06
    symbol
    -0.06
    ブロ
    -0.06
    POSITIVE LOGITS
     crackdown
    0.07
     Assault
    0.07
     �
    0.07
     Attacks
    0.07
     starters
    0.07
     assault
    0.07
     cheering
    0.06
     xpos
    0.06
    تح
    0.06
     LOOK
    0.06
    Activations Density 0.131%

    No Known Activations