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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 58741
    Prev
    Next
    INDEX
    Explanations

    but

    np_max-act · gemini-2.0-flash

    This neuron detects contrastive discourse markers—words like “but,” “however,” etc., that signal an exception or shift in the reasoning.

    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
    っていた
    -0.08
     José
    -0.07
     rpc
    -0.07
     теат
    -0.07
    ुभ
    -0.07
    instrument
    -0.07
     Batman
    -0.07
     дан
    -0.06
     هاي
    -0.06
    ��이
    -0.06
    POSITIVE LOGITS
     MPG
    0.08
     assigns
    0.07
     stays
    0.06
    ụn
    0.06
     but
    0.06
    '''↵↵
    0.06
    _PWM
    0.06
     VARIABLE
    0.05
     ALTER
    0.05
     There
    0.05
    Activations Density 0.073%

    No Known Activations