© 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 EvalsExportsAPI 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. 124677
    Prev
    Next
    INDEX
    Explanations

    beginning of narratives

    np_max-act · gemini-2.0-flash

    This neuron detects role/persona specification tokens — instructions that assign a character identity (names/IDs, ages/century numbers) or first‑person role cues.

    oai_token-act-pair · gpt-5-miniTriggered 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
     stirred
    -0.07
     degrees
    -0.07
    Tur
    -0.07
     LogLevel
    -0.07
     skipped
    -0.07
     bryster
    -0.07
     HIP
    -0.07
     literally
    -0.07
    诜
    -0.07
    appa
    -0.07
    POSITIVE LOGITS
    焼
    0.08
    も多い
    0.07
     YYS
    0.07
    רצח
    0.07
    胺
    0.07
    ច
    0.07
     YM
    0.07
     stronghold
    0.07
    表达了
    0.07
     STORE
    0.06
    Activations Density 0.051%

    No Known Activations