© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 99255
    Prev
    Next
    INDEX
    Explanations

    gradients and variables

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

    This neuron fires on tokens from program or shell output—especially error‐message fragments (like “detached,” “HEAD,” “error”) and other log/trace tokens including numeric & technical outputs.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
    COMPLETE
    -0.71
     Einzel
    -0.70
    რი
    -0.68
     लोक
    -0.66
    hollywood
    -0.66
     ergän
    -0.65
     EQUAL
    -0.65
    ITU
    -0.65
     unsatisfied
    -0.65
     τοῦ
    -0.64
    POSITIVE LOGITS
    requires
    1.27
     Variable
    1.25
     gradients
    1.24
     detach
    1.18
     requires
    1.17
     gradient
    1.12
     detached
    1.10
    Variable
    1.08
    grad
    1.05
     grad
    1.05
    Activations Density 0.010%

    No Known Activations