© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Qwen3-4B
    5. 31-TRANSCODER-HP
    7. 19839
    Prev
    Next
    INDEX
    Explanations

    Technical/academic processes/studies

    np_max-act-logits · gemini-2.0-flash
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_31.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.31.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Negative Logits
    adem
    -0.28
    ama
    -0.25
    loth
    -0.25
    ạm
    -0.25
    quent
    -0.24
    iden
    -0.24
    fi
    -0.24
    redd
    -0.24
    çı°
    -0.24
    <Class
    -0.24
    POSITIVE LOGITS
    agine
    0.35
    纨
    0.30
    orts
    0.30
    å©¿
    0.29
    æĸĩä½ĵ
    0.27
    NST
    0.27
    /tool
    0.26
    NSS
    0.25
    orer
    0.25
     bl
    0.25
    Activations Density 0.033%

    No Known Activations