© 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. 125926
    Prev
    Next
    INDEX
    Explanations

    code/document artifacts

    np_max-act · gemini-2.0-flash

    This neuron detects the boundary token that marks the start of the assistant’s response.

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

    the chat-format boundary that marks the end of a role header and the start of the assistant’s message content.

    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
     баб
    -0.07
    itty
    -0.06
     spelling
    -0.06
    -solving
    -0.06
    uman
    -0.06
    avior
    -0.06
    ollipop
    -0.06
    quisition
    -0.06
    va
    -0.06
    adopt
    -0.06
    POSITIVE LOGITS
    	fflush
    0.07
     продукції
    0.07
    _RAD
    0.06
    !');↵
    0.06
     адміністра
    0.06
     Contents
    0.06
     inclination
    0.06
     asıl
    0.06
    áfico
    0.06
     duro
    0.06
    Activations Density 0.185%

    No Known Activations