© 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 EvalsExportsAPI 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. 6076
    Prev
    Next
    INDEX
    Explanations

    calls to action prompting sign-up

    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
    +0.19
    0.9%
    101
    +0.09
    0.4%
    735
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    592
    +0.19
    0.02
    735
    +0.09
    0.02
    666
    +0.09
    0.02
    Negative Logits
     Himself
    -0.64
     Coul
    -0.61
    geries
    -0.59
     Unloaded
    -0.59
    ãĥīãĥ©ãĤ´ãĥ³
    -0.58
    bler
    -0.58
    ļé
    -0.57
     Galile
    -0.56
    grim
    -0.55
    gery
    -0.55
    POSITIVE LOGITS
     alerts
    0.64
     âĸ
    0.63
     Submit
    0.63
    scribe
    0.63
    ESPN
    0.62
    ername
    0.61
    lish
    0.60
    tainment
    0.59
    riott
    0.59
    ancy
    0.57
    Activations Density 0.024%

    No Known Activations