© 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. 4697
    Prev
    Next
    INDEX
    Explanations

    certain phrases or terms related to news articles and current events

    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
    373
    +0.13
    0.6%
    311
    +0.12
    0.5%
    257
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    683
    +0.13
    0.01
    39
    +0.12
    0.01
    586
    +0.10
    0.01
    Negative Logits
    rosse
    -0.72
    bery
    -0.70
    knit
    -0.68
    nesia
    -0.66
    thro
    -0.61
    ells
    -0.61
     subsequ
    -0.60
     delinquent
    -0.60
    erate
    -0.59
     intellig
    -0.59
    POSITIVE LOGITS
     Shutterstock
    1.09
    utterstock
    0.87
    Magazine
    0.84
     Images
    0.83
     IMAGES
    0.82
    imedia
    0.82
    sburg
    0.76
     Investor
    0.73
    Woman
    0.73
    MSN
    0.72
    Activations Density 0.021%

    No Known Activations