© 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 24576
    9. 8-RES_FS24576-JB
    11. 24024
    Prev
    Next
    INDEX
    Explanations

    commas and conjunctions in complex sentences

    oai_token-act-pair · gpt-4o-miniTriggered 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
    756
    +0.18
    0.8%
    255
    +0.15
    0.7%
    285
    +0.14
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    720
    +0.18
    0.09
    43
    +0.15
    0.07
    285
    +0.14
    0.06
    Negative Logits
     Ranch
    -0.80
    anmar
    -0.75
     Sprint
    -0.72
    SIGN
    -0.68
     Sunny
    -0.66
    TW
    -0.66
    ovember
    -0.66
     Moines
    -0.65
     Anniversary
    -0.65
     dozen
    -0.64
    POSITIVE LOGITS
     whereas
    1.30
     hence
    1.26
     thereby
    1.23
     namely
    1.19
     ie
    1.16
     insofar
    1.13
     thus
    1.12
     consequently
    1.12
     therefore
    1.11
     irrespective
    1.10
    Activations Density 0.440%

    No Known Activations