© 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
    7. Resid Post - 131k
    9. 11-RESID-POST-AA
    11. 12586
    Prev
    Next
    INDEX
    Explanations

    special characters

    np_max-act · gemini-2.0-flash

    The neuron strongly activates on the explicit in‐prompt “respond with …” directives and their bracketed command tokens (e.g. “[”, “end”, “nothing”, “]”), i.e. the formatting markers that specify how the assistant should reply.

    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
    계
    -0.07
    認
    -0.07
    Female
    -0.06
     başkan
    -0.06
    ΙΝ
    -0.06
    720
    -0.06
     بها
    -0.06
    авлива
    -0.06
     aplicación
    -0.06
     наиболее
    -0.06
    POSITIVE LOGITS
    assist
    0.07
     داش
    0.07
     Kod
    0.07
    	Context
    0.06
     Dynasty
    0.06
    _finder
    0.06
    .Amount
    0.06
    _Red
    0.06
     providedIn
    0.06
    abol
    0.06
    Activations Density 0.008%

    No Known Activations