© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 12288
    9. 8-RES_FS12288-JB
    11. 1580
    Prev
    Next
    INDEX
    Explanations

    mentions of statistics or research findings

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    211
    +0.10
    0.4%
    641
    +0.09
    0.4%
    635
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    211
    +0.10
    0.02
    232
    +0.09
    0.02
    343
    +0.09
    0.02
    Negative Logits
    wright
    -0.71
     juggling
    -0.68
    heads
    -0.64
     Ferdinand
    -0.63
     AE
    -0.62
    work
    -0.61
     owl
    -0.61
     Oblivion
    -0.60
     wagon
    -0.60
     directions
    -0.59
    POSITIVE LOGITS
    zanne
    1.34
    arez
    1.21
    pper
    1.18
    ction
    1.17
    icides
    1.16
    icide
    1.15
    icidal
    1.14
    pport
    1.13
    cc
    1.09
    nder
    1.08
    Activations Density 0.045%

    No Known Activations