© 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. 42890
    Prev
    Next
    INDEX
    Explanations

    Referencing entities/things

    np_max-act · gemini-2.0-flash

    The neuron fires on the assistant’s self‐references—tokens like “I,” “assistant,” and other first‐person pronouns or role identifiers.

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

    the neuron detects first-person self-references (speaker's "I"/self‑referential tokens) in the assistant's utterances.

    oai_token-act-pair · gpt-5-miniTriggered 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
     بپ
    -0.07
    bdb
    -0.07
     IndexPath
    -0.07
     контроль
    -0.06
     Est
    -0.06
    icias
    -0.06
     Innoc
    -0.06
    Prefab
    -0.06
     Leak
    -0.06
     pneumonia
    -0.06
    POSITIVE LOGITS
    special
    0.07
    DIFF
    0.07
     closer
    0.07
    .requests
    0.07
     DIFF
    0.06
     perspectives
    0.06
     atroc
    0.06
    нов
    0.06
     labs
    0.06
     �
    0.06
    Activations Density 0.107%

    No Known Activations