© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 5611
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on words and phrases that signal difficulties or obstacles (e.g. “challenge,” “challenges,” “difficulty,” “obstacle”).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_31_width_16k_l0_medium
    Prompts (Dashboard)
    392,802 prompts, 256 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     disgusted
    0.57
     preferring
    0.55
     disgust
    0.53
     risulta
    0.50
    িলারা
    0.50
     autographs
    0.50
     शर्म
    0.49
     делается
    0.49
     prefer
    0.49
     valuables
    0.49
    POSITIVE LOGITS
     challenges
    2.83
    Challenges
    2.72
    challenges
    2.72
     Challenges
    2.66
     desafios
    2.59
     desafíos
    2.44
     challenge
    2.42
    挑战
    2.34
     चुनौतियों
    2.28
     obstacles
    2.27
    Activations Density 0.687%

    No Known Activations