© 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. 2680
    Prev
    Next
    INDEX
    Explanations

    phrases related to the way things are done or the manner in which something is carried out

    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
    341
    +0.12
    0.5%
    576
    +0.10
    0.5%
    46
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    341
    +0.12
    0.03
    576
    +0.10
    0.03
    312
    +0.10
    0.02
    Negative Logits
    oute
    -0.73
    ufact
    -0.72
    oglu
    -0.70
    usters
    -0.68
    fif
    -0.68
    icio
    -0.67
    inently
    -0.62
    alty
    -0.62
    encer
    -0.62
    ospons
    -0.61
    POSITIVE LOGITS
    finding
    0.81
     shape
    0.71
    ward
    0.67
    CHO
    0.67
    WARD
    0.66
    nes
    0.66
     resembles
    0.66
    forward
    0.66
    NE
    0.64
    fare
    0.63
    Activations Density 0.035%

    No Known Activations