© 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
    2. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    3. Llama3.1-8B-IT
    4. Resid Post - 131k
    5. 23-RESID-POST-AA
    6. 29983
    Prev
    Next
    INDEX
    Explanations

    This neuron appears to specialize in detecting **sexually explicit content and NSFW terminology**, particularly focusing on: - **Sexual acts and kinks** (e.g., "breeding," "anal play," "exhibitionism") - **NSFW roleplay dynamics** (e.g., "freeuse," "shared use") - **Slang and vulgar language** tied to erotic contexts (e.g., "bimbo," "creampie") - **Physical descriptors** with sexual connotations (e.g., "wide hips," "full lips") It strongly activates on words and phrases commonly found in adult content, erotic roleplay scenarios, or sexually charged descriptions.

    oai_token-act-pair · deepseek-v3Triggered by @grunklewordner
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_23/trainer_1
    Dataset (Dashboard)
    Various
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.23.hook_resid_post
    Architecture
    standard
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    рап
    -0.07
     cosine
    -0.07
    xor
    -0.07
    еку
    -0.07
    福利
    -0.07
     cozy
    -0.07
    entered
    -0.07
     fundamental
    -0.06
     Parti
    -0.06
     návr
    -0.06
    POSITIVE LOGITS
    (gameObject
    0.06
    дан
    0.06
     bella
    0.06
    _album
    0.06
     О
    0.06
     proceed
    0.06
    ●●●●●●●●
    0.06
    Cancel
    0.06
    \Services
    0.06
    ⟩
    0.06
    Activations Density 0.064%

    No Known Activations