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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 108919
    Prev
    Next
    INDEX
    Explanations

    be

    np_max-act · gemini-2.0-flash

    Same activations, but with all zeros filtered out: <start> replete 0.23706 with 0.32593 includes 0.35425 tell 0.32397 Describe 1.29102 choosing 1.25488 paying 1.14453 special 1.08691 should 1.19336 unique 0.28271 not 0.77099 just 0.57275 cookie 0.27026 cutter 0.20691 description 0.49780 make 0.21655 sure 2.03711 describe 1.30273 keep 0.78613 moving 0.81348 forwards 0.59033 … <end> Explanation of neuron 4 behavior: the main thing this neuron does is find directive or imperative instruction words in the user’s prompt.

    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
     leaks
    -0.07
    around
    -0.07
     evidently
    -0.07
    enc
    -0.07
     unpleasant
    -0.06
     bleak
    -0.06
    MEA
    -0.06
    _by
    -0.06
     Bound
    -0.06
     smokers
    -0.06
    POSITIVE LOGITS
    設備
    0.07
     putas
    0.07
    .pt
    0.06
     εκεί
    0.06
     haar
    0.06
    ्रश
    0.06
    取消
    0.06
    949
    0.06
    學校
    0.06
    .handleChange
    0.06
    Activations Density 0.067%

    No Known Activations