© 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. 126887
    Prev
    Next
    INDEX
    Explanations

    Article previews

    np_max-act · gemini-2.0-flash

    The neuron activates almost exclusively on structural/control tokens (e.g. end-of-text or header markers), i.e. it is detecting metadata/chat formatting tokens rather than natural-language content.

    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
    928
    -0.06
     enc
    -0.06
    Abs
    -0.06
    fb
    -0.06
     FormData
    -0.06
     SUS
    -0.06
     caption
    -0.06
     PUR
    -0.06
    _Reference
    -0.06
    tube
    -0.06
    POSITIVE LOGITS
    _HOUR
    0.07
     هفته
    0.07
     رسانه
    0.07
    적
    0.07
    _escape
    0.07
     mantle
    0.06
    hong
    0.06
    .BorderSize
    0.06
     影
    0.06
    camel
    0.06
    Activations Density 0.008%

    No Known Activations