© 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. 22-TRANSCODER-HP
    7. 88415
    Prev
    Next
    INDEX
    Explanations

    punctuation

    np_max-act · gemini-2.0-flash
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_22.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.22.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    .TAG
    -0.27
    _marshall
    -0.27
    icamente
    -0.27
    :\/\/
    -0.26
    .synthetic
    -0.26
    slaught
    -0.26
    åķĨæĬ¥
    -0.25
    æ®Ĭ
    -0.24
    ahkan
    -0.24
     waged
    -0.24
    POSITIVE LOGITS
     Alfred
    0.27
    Compound
    0.26
    Fe
    0.26
     water
    0.26
     ppt
    0.25
    F
    0.25
    Side
    0.24
    èĤ¯å®ļ
    0.24
     Water
    0.24
    Smart
    0.24
    Activations Density 0.006%

    No Known Activations