© 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. Qwen2.5-7B-IT
    7. Resid Post - 131k
    9. 15-RESID-POST-AA
    11. 4445
    Prev
    Next
    INDEX
    Explanations

    ank

    np_max-act · gemini-2.0-flash

    rhyme-heavy poetic line endings, especially highly assonant, singable endings that anchor verse.

    oai_token-act-pair · gpt-5Triggered by @vetterc0

    end-rhyme candidates, especially words with “-ance/‑ange”-like phonetic endings that are common in poetry.

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    andyrdt/saes-qwen2.5-7b-instruct/resid_post_layer_15/trainer_1
    Dataset (Dashboard)
    Various
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.15.hook_resid_post
    Architecture
    standard
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    إست
    -0.07
    esity
    -0.07
    来袭
    -0.07
     themed
    -0.07
    ifar
    -0.07
     evalu
    -0.06
    WriteBarrier
    -0.06
    窒
    -0.06
    '
    -0.06
    irus
    -0.06
    POSITIVE LOGITS
     Mang
    0.07
     angl
    0.07
     меня
    0.07
    mdl
    0.07
     chants
    0.07
     plants
    0.07
     freelancer
    0.07
    anges
    0.07
     Virginia
    0.06
    än
    0.06
    Activations Density 0.604%

    No Known Activations