© 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. 4842
    Prev
    Next
    INDEX
    Explanations

    mentions of the word "self" at high activations

    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
    635
    +0.12
    0.5%
    497
    +0.10
    0.5%
    323
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    635
    +0.12
    0.03
    323
    +0.10
    0.03
    303
    +0.09
    0.03
    Negative Logits
     XIII
    -0.78
     Nights
    -0.76
     Ashe
    -0.74
    ī
    -0.73
    SHIP
    -0.72
     Rouge
    -0.71
     Syndicate
    -0.69
    dayName
    -0.68
    rium
    -0.67
    IUM
    -0.66
    POSITIVE LOGITS
    destruct
    1.30
    same
    1.02
     destruct
    1.00
    ridges
    0.99
    lessly
    0.98
     explanatory
    0.95
    conscious
    0.92
    contained
    0.91
    ishly
    0.83
    pecially
    0.83
    Activations Density 0.018%

    No Known Activations