© 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
    3. Joseph Bloom · Initial Gemma-2B-IT SAEs
    5. Gemma-2B-IT
    7. Residual Stream
    9. 12-RES-JB
    11. 14328
    Prev
    Next
    INDEX
    Explanations

    superlatives or high-level descriptors, including phrases like "the biggest," "easily the best," and "the most historically active."

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    jbloom/Gemma-2b-IT-Residual-Stream-SAEs/gemma_2b_it_blocks.12.hook_resid_post_16384
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    HuggingFaceFW/fineweb
    Features
    16,384
    Data Type
    float32
    Hook Name
    blocks.12.hook_resid_post
    Hook Layer
    12
    Architecture
    standard
    Context Size
    1,024
    Dataset
    Skylion007/openwebtext
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    674
    +0.10
    0.3%
    1763
    +0.08
    0.2%
    1967
    +0.08
    0.2%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    1522
    +0.10
    0.05
    895
    +0.08
    0.04
    446
    +0.08
    0.04
    Negative Logits
     kasama
    -0.89
     thut
    -0.88
     habang
    -0.86
     kapag
    -0.86
     effe
    -0.81
     coö
    -0.78
     fta
    -0.78
     fto
    -0.75
     fte
    -0.74
     myn
    -0.74
    POSITIVE LOGITS
     most
    0.90
     highest
    0.80
     best
    0.77
     largest
    0.77
     longest
    0.75
     greatest
    0.73
     lowest
    0.72
     finest
    0.72
     biggest
    0.71
     smallest
    0.70
    Activations Density 0.246%

    No Known Activations