Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B-IT
    4. Residual Stream - 131k
    5. 20-GEMMASCOPE-RES-131K
    6. 120761
    Prev
    Next
    INDEX
    Explanations

    words and phrases associated with the quality of being light or uplifting

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k
    Configuration
    google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81
    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
    Not in Any Lists

    No Comments

    Negative Logits
    retudo
    -0.57
    PRNewswire
    -0.57
    enumi
    -0.54
     Ablauf
    -0.52
     AssemblyCompany
    -0.52
     HasFactory
    -0.50
     respeito
    -0.49
    oredCriteria
    -0.49
     steder
    -0.48
    AddTagHelper
    -0.48
    POSITIVE LOGITS
    Light
    1.13
     Light
    1.09
     LIGHT
    1.04
     light
    1.02
    light
    1.01
    LIGHT
    0.91
     Lighter
    0.86
    tralight
    0.81
    lighter
    0.80
     Lightly
    0.79
    Activations Density 0.011%

    No Known Activations