© 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. 13745
    Prev
    Next
    INDEX
    Explanations

    discussions around giveaways, technical glitches, and software functionalities

    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
    763
    +0.12
    0.5%
    55
    +0.10
    0.5%
    393
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    55
    +0.12
    0.05
    102
    +0.10
    0.05
    500
    +0.10
    0.03
    Negative Logits
    ESE
    -0.77
     refrain
    -0.68
    insk
    -0.68
    ievers
    -0.67
    these
    -0.66
    iasco
    -0.63
    few
    -0.63
    uese
    -0.63
    Yan
    -0.63
    635
    -0.61
    POSITIVE LOGITS
     downright
    0.84
    ometric
    0.79
     probably
    0.78
     called
    0.77
     gonna
    0.76
     definitely
    0.75
     undoubtedly
    0.74
     suppose
    0.73
     identical
    0.72
    rontal
    0.69
    Activations Density 0.194%

    No Known Activations