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. Attention Out - 16k
    5. 10-GEMMASCOPE-ATT-16K
    6. 6654
    Prev
    Next
    INDEX
    Explanations

    attends to specific keywords related to failure or error messages from corresponding confirmation or success indicators

    oai_attention-head · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GEMMA-2-2B @ 10-gemmascope-att-16k
    Configuration
    google/gemma-scope-2b-pt-att/layer_10/width_16k/average_l0_70
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.10.attn.hook_z
    Hook Layer
    10
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Head Attr Weights
    0:0.42
    1:0.15
    2:0.07
    3:0.03
    4:0.04
    5:0.06
    6:0.05
    7:0.13
    Negative Logits
     unanim
    -0.31
     BoxDecoration
    -0.30
     obvious
    -0.29
    สำ
    -0.28
     Barnabas
    -0.28
     retic
    -0.28
    inence
    -0.28
     bílé
    -0.28
     azar
    -0.28
    ABILIDAD
    -0.28
    POSITIVE LOGITS
     utafitiHapana
    0.59
    principalTable
    0.49
    ]();
    0.46
    COUVER
    0.42
    ="@+
    0.42
    chtenstein
    0.41
    '));
    
    0.41
     });
    
    0.40
    AndroidJUnit
    0.40
    });
    
    0.40
    Activations Density 0.216%

    No Known Activations