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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 4087
    Prev
    Next
    INDEX
    Explanations

    mentions of uncertainty or concern about future outcomes

    oai_token-act-pair · gpt-3.5-turboTriggered 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
    489
    +0.09
    0.4%
    616
    +0.09
    0.4%
    341
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    271
    +0.09
    0.02
    641
    +0.09
    0.03
    472
    +0.09
    0.02
    Negative Logits
    artney
    -0.68
     Estimates
    -0.65
     OPS
    -0.65
    aples
    -0.64
     Statistical
    -0.62
    Topic
    -0.61
    GP
    -0.60
     Conduct
    -0.58
     Ago
    -0.58
    bard
    -0.57
    POSITIVE LOGITS
    enance
    0.88
     thereof
    0.86
     horr
    0.83
    scape
    0.76
     presented
    0.74
     of
    0.74
    eful
    0.73
     posed
    0.72
    uring
    0.71
    forts
    0.70
    Activations Density 0.069%

    No Known Activations