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. 130427
    Prev
    Next
    INDEX
    Explanations

    could can

    np_max-act · gemini-2.0-flash

    This neuron activates on first-person expressions of internal states, especially “I” followed by verbs like feel, know, can, hear—i.e. self-referential statements of perception or emotion.

    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
    afety
    -0.07
    Nu
    -0.07
    .Usage
    -0.07
    Про
    -0.06
    otyp
    -0.06
     Caul
    -0.06
    黃
    -0.06
    407
    -0.06
    리그
    -0.06
    PLL
    -0.06
    POSITIVE LOGITS
     mote
    0.07
     cả
    0.06
     columna
    0.06
     Prev
    0.06
     Birliği
    0.06
    *'
    0.06
     hauling
    0.06
    ків
    0.06
     capacit
    0.06
     належ
    0.06
    Activations Density 0.010%

    No Known Activations