© 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 24576
    9. 8-RES_FS24576-JB
    11. 5902
    Prev
    Next
    INDEX
    Explanations

    meetings or events happening with high importance or significance

    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
    447
    +0.36
    1.8%
    481
    +0.21
    1.0%
    563
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    563
    +0.36
    0.08
    288
    +0.21
    0.09
    292
    +0.10
    0.06
    Negative Logits
    instead
    -0.65
     *.
    -0.63
    ãģĨ
    -0.63
    mpire
    -0.62
    +(
    -0.59
    .</
    -0.59
     guiActiveUnfocused
    -0.59
    rather
    -0.57
    !.
    -0.57
    Ïī
    -0.56
    POSITIVE LOGITS
     also
    0.91
     concludes
    0.89
     consists
    0.83
     comprises
    0.82
     continues
    0.79
     meanwhile
    0.78
     culmin
    0.77
     began
    0.77
     relies
    0.77
     consisted
    0.76
    Activations Density 0.618%

    No Known Activations