© 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. 35-TRANSCODER-HP
    7. 38626
    Prev
    Next
    INDEX
    Explanations

    say "al" words

    np_max-act-logits · gemini-2.0-flash
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_35.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.35.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Negative Logits
     Constraint
    -0.26
    ç¼¥
    -0.26
    éĢģåĩº
    -0.26
    éĢģåĪ°
    -0.25
     Equality
    -0.24
     weights
    -0.24
    å·²ç»ıè¶ħè¿ĩ
    -0.24
     shorten
    -0.24
    éĻį
    -0.24
    (Error
    -0.24
    POSITIVE LOGITS
    ale
    0.43
    alem
    0.39
    ali
    0.38
    alian
    0.37
    alo
    0.35
    alic
    0.35
    alled
    0.35
    altern
    0.35
    alus
    0.34
    atal
    0.33
    Activations Density 0.375%

    No Known Activations