© 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. 16528
    Prev
    Next
    INDEX
    Explanations

    words related to discussing and explaining various topics or matters

    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
    580
    +0.09
    0.4%
    320
    +0.09
    0.4%
    609
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    348
    +0.09
    0.03
    687
    +0.09
    0.02
    370
    +0.09
    0.02
    Negative Logits
    kai
    -0.79
    İĭ
    -0.74
    aro
    -0.69
    amo
    -0.67
    ois
    -0.66
    falls
    -0.66
    WI
    -0.65
    DragonMagazine
    -0.65
    ennes
    -0.65
    é¾
    -0.65
    POSITIVE LOGITS
     preserving
    0.97
     educating
    0.97
     maximizing
    0.90
     protecting
    0.87
     replacing
    0.82
     exploring
    0.82
     creating
    0.82
     overcoming
    0.81
     promoting
    0.80
     improving
    0.80
    Activations Density 0.062%

    No Known Activations