© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Olmo-3-1125-32B
    3. 32-RES-BATCHTOPK-131K
    4. 115538
    Prev
    Next
    INDEX
    Explanations

    The neuron seems to be detecting citations or references, often marked by a specific character sequence followed by bracketed or parenthesized identifiers. Analyzing the `MAX_ACTIVATING_TOKENS` and `TOKENS_AFTER_MAX_ACTIVATING_TOKEN`, the recurring `Âł` followed by `[` or `[@` strongly suggests this.citations or references like `[@id]`

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    bcywinski/Olmo-3-32B-Base-SAE/saes_allenai_Olmo-3-1125-32B_batch_top_k/resid_post_layer_32
    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
    eren
    -0.07
    aget
    -0.07
    Gratis
    -0.07
    obo
    -0.07
     Fir
    -0.07
    ennes
    -0.07
    otten
    -0.06
    rch
    -0.06
    #index
    -0.06
    pret
    -0.06
    POSITIVE LOGITS
    »
    0.12
    quot
    0.10
    [â̦
    0.10
     jaw
    0.10
     quot
    0.10
    âĢļ
    0.10
     curb
    0.10
     Superv
    0.09
     fr
    0.09
    Č↵
    0.09
    Activations Density 0.116%

    No Known Activations