© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 49152
    9. 8-RES_FS49152-JB
    11. 43070
    Prev
    Next
    INDEX
    Explanations

    variations of the term "as."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    510
    +0.11
    0.5%
    553
    +0.11
    0.5%
    710
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    510
    +0.11
    0.04
    647
    +0.11
    0.04
    192
    +0.10
    0.04
    Negative Logits
     Timbers
    -0.61
    OUS
    -0.61
    erial
    -0.61
    SEE
    -0.60
    aer
    -0.59
    BACK
    -0.57
    KN
    -0.57
    veland
    -0.57
    âĹ¼
    -0.56
     Bringing
    -0.56
    POSITIVE LOGITS
     opposed
    1.07
     usual
    1.03
     evidenced
    0.96
    bestos
    0.85
     advocated
    0.79
     shown
    0.78
    par
    0.77
     advertised
    0.77
    cription
    0.77
    pires
    0.76
    Activations Density 0.089%

    No Known Activations