© 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. 2360
    Prev
    Next
    INDEX
    Explanations

    specific phrases that stand out or hold significance within a broader context

    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
    710
    +0.11
    0.5%
    644
    +0.10
    0.5%
    187
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    644
    +0.11
    0.02
    710
    +0.10
    0.02
    461
    +0.10
    0.01
    Negative Logits
    ÄŁ
    -0.78
    fman
    -0.77
    DERR
    -0.71
     hemor
    -0.71
    llah
    -0.66
    jri
    -0.64
     Indies
    -0.63
     appointments
    -0.62
     Thro
    -0.62
    gur
    -0.61
    POSITIVE LOGITS
    phrase
    1.12
    ology
    1.02
     uttered
    1.00
     phrases
    0.90
     phrase
    0.89
     coined
    0.85
    terday
    0.85
    stress
    0.81
    stick
    0.79
    witz
    0.78
    Activations Density 0.035%

    No Known Activations