Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Andy Arditi · GPT-OSS BatchTopK SAEs
    3. GPT-OSS-20B
    4. Resid Post - 131k
    5. 11-RESID-POST-AA
    6. 112083
    Prev
    Next
    INDEX
    Explanations

    finances

    np_max-act · gemini-2.0-flash

    attends from the start of a document to significant summary statements or informational elements later in the document.

    oai_attention-head · gpt-4oTriggered by @mlfinancebook

    This neuron detects numbers and explicit monetary/financial figures or tokens (amounts, revenues, receipts, taxes, deficits) in financial or budget-related text.

    oai_token-act-pair · gpt-5-miniTriggered by @mlfinancebook
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    andyrdt/saes-gpt-oss-20b/resid_post_layer_11/trainer_0
    Dataset (Dashboard)
    Various
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     principi
    -0.08
     Gru
    -0.08
    няет
    -0.08
     Herv
    -0.08
     Hals
    -0.08
    	location
    -0.07
    ’Etat
    -0.07
     Convention
    -0.07
    �
    -0.07
    	interface
    -0.07
    POSITIVE LOGITS
    收入
    0.14
     revenues
    0.13
     ingresos
    0.11
     revenue
    0.11
     revenus
    0.10
    Revenue
    0.10
     income
    0.10
    .rece
    0.10
     receipts
    0.10
     receitas
    0.10
    Activations Density 0.028%

    No Known Activations