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. 18-GEMMASCOPE-RES-16K
    6. 8737
    Prev
    Next
    INDEX
    Explanations

    expressions indicating human emotions and relationships

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

    Following verbs, implying direction/turning

    np_acts-logits-general · gemini-2.0-flash

    avoidance and turning away

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k
    Configuration
    google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.18.hook_resid_post
    Hook Layer
    18
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     AssemblyProduct
    -0.74
    Autoritní
    -0.74
    ArgsConstructor
    -0.67
     الحره
    -0.60
     ExecuteAsync
    -0.60
     SEDS
    -0.59
     Familienname
    -0.58
    WriteTagHelper
    -0.57
    awtextra
    -0.56
    RenderAtEndOf
    -0.56
    POSITIVE LOGITS
     avoidance
    1.54
     avoid
    1.41
     avoiding
    1.38
     avoided
    1.35
    avoid
    1.32
     AVOID
    1.30
    Avoid
    1.30
     Avoid
    1.29
     Avoidance
    1.28
     avoids
    1.28
    Activations Density 0.503%

    No Known Activations