© 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. Qwen3-4B
    5. 32-TRANSCODER-HP
    7. 42910
    Prev
    Next
    INDEX
    Explanations

    non-English

    np_max-act-logits · gemini-2.0-flash
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_32.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.32.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     vá»įng
    -0.29
    lauf
    -0.28
    ussen
    -0.28
    rib
    -0.28
    æ´¾
    -0.27
    uos
    -0.27
    妥
    -0.26
    rub
    -0.26
    /library
    -0.26
    fill
    -0.26
    POSITIVE LOGITS
    æĺĮ
    0.28
     softmax
    0.27
    转移åĪ°
    0.25
    è°Į
    0.25
    éĢĥçĶŁ
    0.24
     mcc
    0.24
    å¾ĹåĪĨ
    0.24
    еÑĪ
    0.23
    MF
    0.23
    (interval
    0.23
    Activations Density 0.006%

    No Known Activations