© 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. 14146
    Prev
    Next
    INDEX
    Explanations

    Avoid

    np_max-act · gemini-2.0-flash

    The neuron activates on imperative “avoid” (or its translations) instructions—that is, commands telling the reader to refrain from or avoid something.

    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
    けて
    -0.07
     leur
    -0.07
     sweetheart
    -0.06
    -opening
    -0.06
     cleaned
    -0.06
    .isNotBlank
    -0.06
     SCR
    -0.06
    electric
    -0.06
     expiration
    -0.06
    tearDown
    -0.06
    POSITIVE LOGITS
    aty
    0.06
    mut
    0.06
    .bs
    0.06
    usb
    0.06
    Candidate
    0.06
    -alpha
    0.06
    beeld
    0.06
     tts
    0.06
     موس
    0.06
     adolescent
    0.06
    Activations Density 0.012%

    No Known Activations