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. 393
    Prev
    Next
    INDEX
    Explanations

    selection and probability

    np_max-act · gemini-2.0-flash

    attends from higher-numbered types to lower-numbered types in ordered sequences or items, such as numeric identifiers or item codes, with consistent specificity where the higher-numbered type acts as a more general class or sequence index.

    oai_attention-head · gpt-4oTriggered 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
     Largo
    -0.08
    apun
    -0.08
    ffic
    -0.08
    mf
    -0.07
    frica
    -0.07
     Anwendungen
    -0.07
     Everywhere
    -0.07
     countless
    -0.07
    "↵/
    -0.07
    astian
    -0.07
    POSITIVE LOGITS
     sélection
    0.09
     selección
    0.09
     criterios
    0.08
     shards
    0.08
     selectie
    0.08
     exposure
    0.08
     composition
    0.08
    .person
    0.08
     Composition
    0.08
     drawn
    0.08
    Activations Density 0.040%

    No Known Activations