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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-2B
    4. Residual Stream - 16k
    5. 22-GEMMASCOPE-RES-16K
    6. 3559
    Prev
    Next
    INDEX
    Explanations

    significant performance differences in test outcomes

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

    "More" or "less" comparisons

    np_acts-logits-general · gemini-2.0-flash

    comparison words following performance or quantity

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 22-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_22/width_16k/average_l0_72
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.22.hook_resid_post
    Hook Layer
    22
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    much
    -0.64
     znacznie
    -0.59
    mucho
    -0.56
    very
    -0.55
     paljon
    -0.55
     намного
    -0.54
    훨
    -0.53
    Fully
    -0.51
    VERY
    -0.49
     dobry
    -0.48
    POSITIVE LOGITS
    rungsseite
    1.11
    RegressionTest
    0.88
     ujednoznacz
    0.83
    曖昧さ回避
    0.80
     resourceCulture
    0.77
    verwijspagina
    0.77
    aarrggbb
    0.76
    findpost
    0.74
    出版年
    0.71
    principalTable
    0.70
    Activations Density 0.409%

    No Known Activations