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. 86146
    Prev
    Next
    INDEX
    Explanations

    to

    np_max-act · gemini-2.0-flash

    The neuron flags action-oriented technical terms—particularly verbs and nouns that describe operations or commands (e.g. “write,” “image,” “torque,” “control,” “stop”) in procedural/instructional contexts.

    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
    sku
    -0.07
    iedo
    -0.07
     Shield
    -0.06
     budou
    -0.06
    rie
    -0.06
     Ri
    -0.06
     gab
    -0.06
    ansa
    -0.06
     eps
    -0.06
    <!--↵
    -0.06
    POSITIVE LOGITS
     وب
    0.07
     sequelize
    0.06
    akra
    0.06
     Contr
    0.06
     指
    0.06
    [section
    0.06
     процесса
    0.06
     parameters
    0.06
    antically
    0.06
     dispenser
    0.06
    Activations Density 0.222%

    No Known Activations