© 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
    Side-by-side compare needs a wider screen.
    Open other featureClose compare
    PrevNext
    Close
    Circuit Sparsity
    No connected neurons data available
    Explanations

    `self.heap_array` or `_insert`- `i`: `if`, `import`, variable `i`- `st`: `self.ancestor`, `list`, `str`- `ro`: `self.root`- `m`: `self.heapify_down`, `m.root`- `h`: `self.heapify_down`, `self.parent(i)`, `while`, `def` or recursive calls `a(h)`Let's look at them as part of keywords or identifiers related to data structures.- `self.heap_array`- `self.heapify_up`- `self.heapify_down`- `self.root`- `def h`- `def i`The `TOKENS_AFTER_MAX_ACTIVATING_TOKEN` are: `array`, `x`, `or`, `ot`, `ot`, `.`, `=`, `)`, `)`, `)`Context of `ro`: `self. ★ro★ot = f` or `self. ★ro★ot = a(b)`. The token after `ro` is `ot`. Together it forms `root`.Context of `h`: `a(★h★)`. The token after `h` is `)`. This suggests function calls or variable accesses like `a(h)`.Context of `m`: `★m★.root.right.right = a('str')`. Token after `m` is `.`. This is accessing an attribute.The majority of the `MAX_ACTIVATING_TOKENS` and their surrounding context point towards common variable names and identifiers within Python classes, particularly for data structures like heaps and trees.`self.heap_array`, `self.root`, `self.ancestor`, `a(h)` (recursive call argument), `m.root` (object attribute access).The pattern `self.` followed by an identifier, or common function/variable names `root`, `heap_array`, `ancestor`, `h`, or `m` for object instances."self" attribute access or common identifiers

    python-code · gemini-2.5-flash-lite

    list element assignment (negative activations)

    python-code · gemini-2.5-flash-lite
    Auto-Interp Tools
    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
    Negative Logits
     Ex
    -0.02
     
    -0.02
    ases
    -0.02
     exp
    -0.01
    `
    -0.01
    tic
    -0.01
    ates
    -0.01
    z
    -0.01
    j
    -0.01
     ->
    -0.01
    Positive Logits
    s
    0.04
     enum
    0.03
    rip
    0.03
    ateg
    0.03
    ¢
    0.03
    den
    0.03
    en
    0.03
    sen
    0.03
    ren
    0.03
    ess
    0.02
    Activations Density 100.000%

    No Known Activations

    PrevNext
    Close
    Circuit Sparsity
    No connected neurons data available
    Explanations
    No Explanations Found
    Auto-Interp Tools
    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
    Negative Logits
    &
    0.00
    *
    0.00
    #
    0.00
    '
    0.00
    )
    0.00
    %
    0.00
    !
    0.00
    "
    0.00
    $
    0.00
    (
    0.00
    Positive Logits
    &
    0.00
    *
    0.00
    #
    0.00
    '
    0.00
    )
    0.00
    %
    0.00
    !
    0.00
    "
    0.00
    $
    0.00
    (
    0.00
    Activations Density 100.000%

    No Known Activations