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
    4. Residual Stream - 131k
    5. 20-GEMMASCOPE-RES-131K
    6. 53894
    Prev
    Next
    INDEX
    Explanations

    specific technical or procedural details related to research methodologies

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    Common prepositions or function words (by, from, of, to, with, etc.) that connect elements in academic, technical or formal writing contexts.

    eleuther_acts_top20 · claude-3-7-sonnet-20250219Triggered 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
    Not in Any Lists

    No Comments

    Negative Logits
     itſelf
    -0.48
     whoſe
    -0.46
    ConstraintMaker
    -0.45
    isible
    -0.44
    bitField
    -0.43
    ſch
    -0.43
     appartiennent
    -0.43
     ſhould
    -0.42
    CAL
    -0.42
     يتيمه
    -0.40
    POSITIVE LOGITS
     propOrder
    0.54
    ########.
    0.54
    Personendaten
    0.51
    хьтан
    0.41
    __':
    
    0.40
     surla
    0.39
    DotNetBar
    0.39
    __':
    0.36
    typeorm
    0.36
     RIT
    0.36
    Activations Density 0.305%

    No Known Activations