© 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. 124587
    Prev
    Next
    INDEX
    Explanations

    code, software documentation

    np_max-act · gemini-2.0-flash

    The neuron activates on the special “assistant” role‐label token, effectively detecting the start of an assistant’s message.

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

    names of famous literary works from Russian literature.

    oai_token-act-pair · gpt-4o-miniTriggered by @xinyanhu8

    markers for speaker headers in chat transcripts, especially the assistant role within conversation 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
     coords
    -0.07
    ANG
    -0.06
     hats
    -0.06
    Cli
    -0.06
     Naturally
    -0.06
     weil
    -0.06
     fakat
    -0.06
    stackpath
    -0.06
     quiero
    -0.06
    	await
    -0.06
    POSITIVE LOGITS
    ète
    0.07
     storyboard
    0.07
    ıyor
    0.06
    _taxonomy
    0.06
     geometry
    0.06
     Taking
    0.06
     trab
    0.06
    .defer
    0.06
     ấy
    0.06
    -connected
    0.06
    Activations Density 0.187%

    No Known Activations