© 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. 30-TRANSCODER-HP
    7. 96592
    Prev
    Next
    INDEX
    Explanations

    dismantling, dissecting, or demolishing

    np_max-act-logits · gemini-2.0-flash
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_30.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.30.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    swift
    -0.28
     am
    -0.28
     tid
    -0.26
    zá
    -0.26
    IZED
    -0.25
    dar
    -0.25
    ëĶĺ
    -0.25
    chor
    -0.25
    ÑĤÑģÑı
    -0.25
     corre
    -0.25
    POSITIVE LOGITS
    æĭĨéĻ¤
    0.31
    æ¯ģçģŃ
    0.28
     actionTypes
    0.27
    ispers
    0.27
    æĭĨ
    0.26
    åĵĪå°Ķ
    0.26
    YLES
    0.24
    	Destroy
    0.24
     unpredict
    0.24
     Infant
    0.24
    Activations Density 0.007%

    No Known Activations