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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    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. 8357
    Prev
    Next
    INDEX
    Explanations

    technical terms and concepts related to data processing and programming

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

    followed by modal verbs

    np_acts-logits-general · gemini-2.0-flash

    would then

    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
    GenerationType
    -0.84
    ]--;
    -0.84
    __":
    
    -0.84
     CWE
    -0.80
     propOrder
    -0.77
    __':
    
    -0.76
    ]='\
    -0.71
    σθαι
    -0.71
    "])
    
    -0.71
    endix
    -0.70
    POSITIVE LOGITS
     would
    0.77
    would
    0.63
    わけです
    0.63
     wouldn
    0.63
     WOULD
    0.59
     because
    0.58
    Would
    0.58
     Would
    0.57
     puisqu
    0.57
     =
    0.56
    Activations Density 0.911%

    No Known Activations