© 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
    2. Qwen3.5-4B
    3. 15-RES-MATRYOSHKA-65K
    4. 19535
    Prev
    Next
    INDEX
    Explanations

    The neuron seems to detect programming variable names followed by common syntax elements. Let's look for a concise phrase capturing this.- **MAX_ACTIVATING_TOKENS**: `product`, `output`, `counter`, `number`, `tree`, `child`. These are variable-like identifiers.- **TOKENS_AFTER_MAX_ACTIVATING_TOKEN**: `);`, `)`, `value`, `)){`, `body`, `->`, `;`. These are syntax elements commonly following identifiers.- **TOP_ACTIVATING_TEXTS**: Contain numerous code snippets showing these `identifier.syntax` or `identifier(args)` patterns.The neuron is primarily associated with code structure involving identifiers and their immediate follow-up syntax.The explanation needs to be 3-20 words, concise, and descriptive of what it detects/predicts.It finds patterns in lists.Looking for patterns in programming code snippets.programming syntax after identifiers

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    decoderesearch/qwen-3.5-saes/qwen-3.5-4b
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    æĪIJä¸Ģ
    -0.06
     ailleurs
    -0.05
    æĸIJ
    -0.05
     posit
    -0.05
    æ®
    -0.05
    order
    -0.05
    æĪIJçļĦ
    -0.05
    æ¶ī
    -0.05
    çĬ¶
    -0.05
    ine
    -0.05
    POSITIVE LOGITS
    enj
    0.07
     tribes
    0.06
     tribe
    0.06
    꽤
    0.06
    æľ¬é¢Ĩ
    0.06
     frigo
    0.06
    ãģĮ決
    0.06
    emie
    0.06
    asons
    0.06
    emic
    0.06
    Activations Density 0.023%

    No Known Activations