© 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. 20974
    Prev
    Next
    INDEX
    Explanations

    web links and dates in a particular format

    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
    314
    +0.10
    0.4%
    529
    +0.09
    0.4%
    399
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    399
    +0.10
    0.02
    6
    +0.09
    0.02
    100
    +0.09
    0.02
    Negative Logits
    entimes
    -0.84
     Lomb
    -0.81
     suspic
    -0.68
     pockets
    -0.68
     manif
    -0.65
     spilling
    -0.65
     infringing
    -0.64
    antha
    -0.63
     outsiders
    -0.63
    hiba
    -0.62
    POSITIVE LOGITS
     âĢİ
    0.82
     Appears
    0.82
    ãĥĺ
    0.79
    BUT
    0.77
    SIGN
    0.76
     Directive
    0.76
    prop
    0.72
    =#
    0.71
    Done
    0.70
    relations
    0.69
    Activations Density 0.027%

    No Known Activations