© 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. 5181
    Prev
    Next
    INDEX
    Explanations

    discussions about political and governmental 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
    447
    +0.19
    0.9%
    640
    +0.14
    0.6%
    679
    +0.13
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    6
    +0.19
    0.09
    679
    +0.14
    0.09
    440
    +0.13
    0.07
    Negative Logits
    cipled
    -0.89
     affili
    -0.77
     workspace
    -0.69
     overturn
    -0.68
     strugg
    -0.67
     interpreter
    -0.64
     cabbage
    -0.64
    sth
    -0.62
     summ
    -0.61
    oing
    -0.61
    POSITIVE LOGITS
     Specifically
    1.32
     Whereas
    1.31
     Whilst
    1.26
     Firstly
    1.18
     While
    1.18
     Normally
    1.18
     Suppose
    1.17
     Luckily
    1.17
     Again
    1.16
     Thankfully
    1.14
    Activations Density 0.359%

    No Known Activations