© 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
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 61161
    Prev
    Next
    INDEX
    Explanations

    hard + noun

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

    The neuron activates on occurrences of the word “hard” in noun-modifier compounds (e.g. hard copy, hard rock, hard drug, hard case, etc.).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
    </h3>
    -2.36
    </sub>
    -1.77
    ſelves
    -1.74
    噥
    -1.73
    でございます
    -1.73
    菹
    -1.73
    ruguay
    -1.66
     RECOMMEND
    -1.65
    -1.62
     There
    -1.61
    POSITIVE LOGITS
     hard
    2.50
    ции
    2.03
    Hard
    1.98
    hard
    1.96
    ！”
    1.89
     Hard
    1.83
     membag
    1.73
    。”
    1.70
    an
    1.66
    3
    1.61
    Activations Density 0.012%

    No Known Activations