© 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 · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 12288
    9. 8-RES_FS12288-JB
    11. 9865
    Prev
    Next
    INDEX
    Explanations

    the word "so" when activated

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    Data Type
    torch.float32
    Hook Point
    blocks.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    567
    +0.10
    0.4%
    699
    +0.09
    0.4%
    509
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    509
    +0.10
    0.02
    699
    +0.09
    0.02
    540
    +0.09
    0.02
    Negative Logits
     Flavoring
    -0.65
    aughed
    -0.62
     Lau
    -0.62
    ª
    -0.59
     Neighborhood
    -0.59
     Kang
    -0.59
     Dreams
    -0.59
     Hunters
    -0.58
    ²
    -0.58
     Fallen
    -0.57
    POSITIVE LOGITS
    oner
    1.04
     anonymously
    0.92
    zin
    0.86
     cheaply
    0.84
     responsibly
    0.82
     willingly
    0.80
     voluntarily
    0.79
    othe
    0.79
    ----------------------------------------------------------------
    0.78
     reluctantly
    0.76
    Activations Density 0.043%

    No Known Activations