© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. OpenAI · Weight-Sparse Transformers
    5. CircuitGPT-Python
    7. Residual Stream
    9. 7-RESID
    11. 1828
    Prev
    Next
    INDEX
    Explanations

    attribute access with underscore or dot (negative activations)

    python-code · gemini-2.5-flash-lite

    second part of `base64.b64decode`

    python-code · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    Prompts (Dashboard)
    16,384 prompts, 256 tokens each
    Dataset (Dashboard)
    neuronpedia/python-code-simplified
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    work
    -0.36
    ork
    -0.18
     var
    -0.14
    straint
    -0.12
    ariable
    -0.08
    si
    -0.08
    str
    -0.07
    straints
    -0.07
    26
    -0.05
    11
    -0.03
    POSITIVE LOGITS
    xp
    0.23
    tex
    0.22
    mport
    0.22
    ft
    0.22
    fset
    0.21
    64
    0.17
    ink
    0.16
    rc
    0.16
    mbol
    0.16
    _exp
    0.15
    Activations Density 99.997%

    No Known Activations