© 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. 5417
    Prev
    Next
    INDEX
    Explanations

    phrases related to sensitive, confidential information and speaking on the condition of anonymity

    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
    373
    +0.14
    0.6%
    317
    +0.09
    0.4%
    52
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    317
    +0.14
    0.01
    52
    +0.09
    0.01
    214
    +0.09
    0.01
    Negative Logits
    ils
    -0.69
    issance
    -0.68
    semble
    -0.64
    vantage
    -0.64
     Silver
    -0.63
    Dust
    -0.63
    kil
    -0.61
     Trilogy
    -0.61
    clipse
    -0.60
    ammy
    -0.58
    POSITIVE LOGITS
    aldehyde
    0.74
     permission
    0.70
     authorization
    0.68
     confir
    0.67
     deletion
    0.66
    taker
    0.65
    ality
    0.65
    eering
    0.64
     clauses
    0.63
     assurances
    0.62
    Activations Density 0.020%

    No Known Activations