© 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
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 24576
    9. 8-RES_FS24576-JB
    11. 24167
    Prev
    Next
    INDEX
    Explanations

    instances of the phrase "in case."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    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
    535
    +0.16
    0.8%
    8
    +0.10
    0.4%
    19
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    535
    +0.16
    0.02
    348
    +0.10
    0.02
    599
    +0.10
    0.01
    Negative Logits
    elta
    -0.81
    vez
    -0.74
    opter
    -0.73
    ieved
    -0.73
    reath
    -0.72
    epend
    -0.71
    pees
    -0.70
    enfranch
    -0.70
    pez
    -0.69
    Rated
    -0.69
    POSITIVE LOGITS
     you
    0.82
     anyone
    0.81
     anybody
    0.76
     somebody
    0.69
     someone
    0.69
     they
    0.68
     ever
    0.68
     else
    0.66
     we
    0.66
     guests
    0.66
    Activations Density 0.023%

    No Known Activations