© 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. 22439
    Prev
    Next
    INDEX
    Explanations

    terms related to rarity and classification in various contexts

    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
    373
    +0.20
    0.9%
    407
    +0.10
    0.5%
    145
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    176
    +0.20
    0.03
    326
    +0.10
    0.04
    316
    +0.10
    0.03
    Negative Logits
     respons
    -0.71
     charism
    -0.65
     behav
    -0.64
     inherently
    -0.62
     spokes
    -0.61
     enthus
    -0.61
     scr
    -0.61
     distinctive
    -0.61
     polarized
    -0.61
     vibr
    -0.60
    POSITIVE LOGITS
    pees
    0.70
     698
    0.65
    chieve
    0.64
    ificant
    0.63
    payer
    0.62
    anton
    0.62
    osterone
    0.61
    abytes
    0.61
     Deaths
    0.60
    usercontent
    0.59
    Activations Density 0.151%

    No Known Activations