© 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. Qwen2.5-7B-IT
    7. Resid Post - 131k
    9. 15-RESID-POST-AA
    11. 122835
    Prev
    Next
    INDEX
    Explanations

    Code/Random characters

    np_max-act · gemini-2.0-flash

    This neuron activates on tokens that are part of the assistant's generated replies (i.e., it detects assistant-role output).

    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
    kę
    -0.08
    解决了
    -0.07
    Seven
    -0.07
     prep
    -0.07
    	now
    -0.07
     *
↵
    -0.07
    קום
    -0.07
    Dos
    -0.07
    America
    -0.07
    Mrs
    -0.07
    POSITIVE LOGITS
    𫚭
    0.08
    _height
    0.07
    radient
    0.07
    кладыва
    0.07
    ♏
    0.07
     מפתח
    0.07
     Structural
    0.07
    imb
    0.07
    .OrderByDescending
    0.06
    .Ribbon
    0.06
    Activations Density 0.075%

    No Known Activations