Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    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. 109612
    Prev
    Next
    INDEX
    Explanations

    out

    np_max-act · gemini-2.0-flash

    This neuron selectively activates on the verb “carried” (as in “carried out”), i.e. it detects that passive‐voice procedural phrasing.

    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
    agina
    -0.07
     pick
    -0.07
    áhnout
    -0.07
     Gree
    -0.07
     Nice
    -0.07
     picks
    -0.06
     bombs
    -0.06
     Nation
    -0.06
     sneak
    -0.06
    丝
    -0.06
    POSITIVE LOGITS
    Run
    0.09
     الو
    0.08
     Continuous
    0.07
    -в
    0.07
     Coun
    0.07
    320
    0.07
     Run
    0.07
    30
    0.07
    роб
    0.07
    htdocs
    0.06
    Activations Density 0.005%

    No Known Activations