© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    5. Llama3.1-8B-IT (Instruct)
    7. Resid Post - 131k
    9. 11-RESID-POST-AA
    11. 42648
    Prev
    Next
    INDEX
    Explanations

    Beginning of document

    np_max-act · gemini-2.0-flash

    This neuron activates on uppercase section headings and keywords (e.g., “SHORT TITLE,” “FINDINGS,” “ALLOWANCE,” “INCOME,” “REDUCTION”), so it’s looking for block-style section titles or headings in all caps.

    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
    dığı
    -0.08
     Buff
    -0.07
     Sanders
    -0.07
     Sword
    -0.06
    edBy
    -0.06
    umeric
    -0.06
     Ward
    -0.06
    ContentSize
    -0.06
    burgh
    -0.06
    уг
    -0.06
    POSITIVE LOGITS
     �
    0.07
     decentral
    0.06
    IllegalAccessException
    0.06
     investigates
    0.06
    covered
    0.06
     triang
    0.06
     refers
    0.06
    .setLocation
    0.06
    (chunk
    0.06
     dateTime
    0.05
    Activations Density 0.002%

    No Known Activations