© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    5. Gemma-2-9B
    7. Residual Stream - 131k
    9. 20-GEMMASCOPE-RES-131K
    11. 105304
    Prev
    Next
    INDEX
    Explanations

    references to temporal sequences and specific points in time

    oai_token-act-pair · gpt-4o-mini
    Triggered by @bot

    Common connecting words or phrases that introduce clauses or establish relationships between ideas, particularly temporal markers ("at"), conditional qualifiers ("but", "though", "would"), knowledge indicators ("know", "know my"), or comparative phrases ("is", "are", "definitely allocating three", "not it", "in part", "at least in part").

    eleuther_acts_top20 · claude-3-7-sonnet-20250219
    Triggered by @lu-christina
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.20.hook_resid_post
    Hook Layer
    20
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Negative Logits
    цездатний
    -0.60
    kloped
    -0.58
     للاسماء
    -0.52
    ConstraintMaker
    -0.50
     Geſch
    -0.48
     oprot
    -0.48
    المكان
    -0.47
     Exacts
    -0.47
     Verſ
    -0.46
     Houſe
    -0.46
    POSITIVE LOGITS
    少なくとも
    0.40
     onCancelled
    0.36
    ReusableCell
    0.34
    ፩
    0.32
    FailureListener
    0.31
    getComponent
    0.31
    AddTagHelper
    0.31
     sure
    0.31
    我知道
    0.30
     definitely
    0.29
    Activations Density 0.201%

    No Known Activations