© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 58318
    Prev
    Next
    INDEX
    Explanations

    y

    np_max-act · gemini-2.0-flash

    The neuron fires on the special “header” or speaker‐label tokens (e.g. the <|start_header_id|>, <|end_header_id|>, “assistant,” etc.) that mark conversation metadata rather than on normal text.

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

    markers indicating the start of an assistant’s turn in the chat-style header/metadata.

    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
    Subjects
    -0.07
    -----------*/↵
    -0.07
     farm
    -0.06
     ↵    ↵
    -0.06
    JOB
    -0.06
    throp
    -0.06
     //}↵↵
    -0.06
     Committee
    -0.06
     committees
    -0.06
    ID
    -0.06
    POSITIVE LOGITS
     движ
    0.07
    .LookAndFeel
    0.07
     возду
    0.07
    -animation
    0.06
     UIG
    0.06
    犬
    0.06
    _spawn
    0.06
    .Automation
    0.06
    _CONST
    0.06
     дж
    0.06
    Activations Density 0.046%

    No Known Activations