© 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. 70909
    Prev
    Next
    INDEX
    Explanations

    This neuron detects instances of the present‐perfect construction (forms of “have/has” plus a past participle).

    oai_token-act-pair · o4-miniTriggered by @msahalmulki

    indirectly reported statements and claims (the content of what someone said/claimed/maintained), especially in explanatory or defensive attributions using pronouns and auxiliaries.

    oai_token-act-pair · gpt-5Triggered by @msahalmulki
    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
     vectors
    -0.08
    Med
    -0.08
    Der
    -0.07
    мон
    -0.07
    -esteem
    -0.07
    swift
    -0.07
    澤
    -0.07
    嶝
    -0.06
    drs
    -0.06
    cel
    -0.06
    POSITIVE LOGITS
    时光
    0.09
    衣物
    0.08
    希望自己
    0.07
    _topic
    0.07
    싶
    0.07
     boon
    0.07
     Expansion
    0.07
     optimism
    0.06
     BI
    0.06
    言われ
    0.06
    Activations Density 0.076%

    No Known Activations