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. 90022
    Prev
    Next
    INDEX
    Explanations

    instances of significant actions or decisions made in various contexts

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

    The highlighted tokens frequently appear at the end of noun phrases that represent physical locations, objects, or concepts that can contain or hold something (e.g., "in the media," "at metamorphosis," "in the warehouses," "source control," "high probability").

    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
    下载附件
    -0.41
    verwijspagina
    -0.31
     puissiez
    -0.30
     décé
    -0.30
     vapa
    -0.29
     régimen
    -0.29
    __':
    -0.29
    ฬ
    -0.28
    trica
    -0.28
     Moscú
    -0.28
    POSITIVE LOGITS
    ViewFeatures
    0.59
     FetchType
    0.57
    AxisAlignment
    0.56
    fromnode
    0.54
     newBuilder
    0.53
     يتيمه
    0.53
    /*++
    0.52
    OGND
    0.51
    MENAFN
    0.51
     Italijanski
    0.51
    Activations Density 0.678%

    No Known Activations