© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Google DeepMind · Exploring Gemma 2 with Gemma Scope
    3. Gemma-2-9B
    4. Residual Stream - 16k
    5. 28-GEMMASCOPE-RES-16K
    6. 7094
    Prev
    Next
    INDEX
    Explanations

    phrases that express necessity or requirements for improvement

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

    Comparison/need for improvement

    np_acts-logits-general · gemini-2.0-flash

    more and better

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-9B @ 28-gemmascope-res-16k
    Configuration
    google/gemma-scope-9b-pt-res/layer_28/width_16k/average_l0_119
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.28.hook_resid_post
    Hook Layer
    28
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    most
    -0.34
    MOST
    -0.30
     cał
    -0.28
    roppo
    -0.28
     sisa
    -0.27
    autaire
    -0.27
    heu
    -0.27
    uesta
    -0.26
     vừa
    -0.26
     Ray
    -0.26
    POSITIVE LOGITS
    もう少し
    1.13
    もっと
    1.11
     more
    1.08
     lebih
    0.91
     stronger
    0.90
     better
    0.88
     clearer
    0.87
     více
    0.86
    更多
    0.85
     mehr
    0.84
    Activations Density 0.364%

    No Known Activations