© 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. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 67394
    Prev
    Next
    INDEX
    Explanations

    seen many times

    np_acts-logits-general · gemini-2.5-flash-lite

    The neuron is picking out first‐person narrative cues—tokens like “I,” “I’m,” “I’ve,” and related personal-experience verbs—marking where the author speaks in the first person.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ↵↵↵
    -1.40
     =
    -1.40
     similar
    -1.18
    ↵↵↵↵
    -1.13
    </h4>
    -1.09
     customer
    -1.07
     filter
    -1.05
    =
    -1.05
    .
    -1.05
     >
    -1.03
    POSITIVE LOGITS
     pommes
    1.56
     berge
    1.49
     radikal
    1.46
     vanille
    1.42
     perpé
    1.38
     spesial
    1.30
     intelek
    1.29
     vaksin
    1.28
     nôtre
    1.28
     corrom
    1.26
    Activations Density 0.029%

    No Known Activations