© 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. 128873
    Prev
    Next
    INDEX
    Explanations

    later

    np_max-act · gemini-2.0-flash

    The neuron is triggered by tokens naming specific years or numeric dates and geographic/place names (e.g. “1973,” “Prague,” “Germany”).

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

    encyclopedic, proper‑noun–dense expository text (biographies, scientific/technical summaries, and historical/event write‑ups with dates and institutions).

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    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
    .De
    -0.06
    (CG
    -0.06
     Model
    -0.06
     маль
    -0.06
     Between
    -0.06
     ALS
    -0.05
    aliz
    -0.05
     derog
    -0.05
     Deprecated
    -0.05
     humili
    -0.05
    POSITIVE LOGITS
     сах
    0.07
     splendid
    0.07
     졸
    0.07
    众
    0.07
    ِه
    0.07
    RSpec
    0.07
    久
    0.07
    _write
    0.07
    เมตร
    0.07
    bine
    0.06
    Activations Density 15.432%

    No Known Activations