Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    3. Llama3.1-8B-IT (Instruct)
    4. Resid Post - 131k
    5. 11-RESID-POST-AA
    6. 130200
    Prev
    Next
    INDEX
    Explanations

    rocess

    np_max-act · gemini-2.0-flash

    This neuron detects instances of the “re-” prefix on verbs (e.g. repackage, reapply, reprocess, repost, rephrase, redesign, repaint), signaling an action of doing something again.

    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
    <Vector
    -0.07
     نمی
    -0.07
    porno
    -0.06
     несколько
    -0.06
     offre
    -0.06
     เช
    -0.06
     honorary
    -0.06
     wrongdoing
    -0.06
     önemlidir
    -0.05
     scarcely
    -0.05
    POSITIVE LOGITS
     relocated
    0.09
     rebuild
    0.09
     redesigned
    0.09
     resett
    0.08
     rebuilt
    0.08
     rebuilding
    0.08
     redefine
    0.07
     resale
    0.07
     repeal
    0.07
    plot
    0.07
    Activations Density 0.074%

    No Known Activations