© 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. 124733
    Prev
    Next
    INDEX
    Explanations

    scams or urgency

    np_max-act · gemini-2.0-flash

    The neuron detects words and tokens expressing persuasive or high-pressure tactics—e.g. urging immediacy, convincing, pressuring, claiming—to coerce the reader.

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

    phrases that create urgency or high-pressure demands to act immediately (time-limited threats/pressure tactics).

    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
     nop
    -0.08
    Mode
    -0.07
    Ross
    -0.07
    PRE
    -0.06
     fares
    -0.06
     Ged
    -0.06
     KING
    -0.06
    eroon
    -0.06
     Liam
    -0.06
     Çağ
    -0.06
    POSITIVE LOGITS
     %(
    0.08
    _documents
    0.07
    (\$
    0.07
    _undo
    0.07
     Nimbus
    0.06
    .UPDATE
    0.06
     Mosul
    0.06
    'nda
    0.06
    _instance
    0.06
     "${
    0.06
    Activations Density 0.123%

    No Known Activations