© 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. 18276
    Prev
    Next
    INDEX
    Explanations

    the word "hence" followed by supporting reasons or conclusions

    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
    709
    +0.12
    0.6%
    334
    +0.10
    0.4%
    77
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    334
    +0.12
    0.02
    155
    +0.10
    0.01
    32
    +0.09
    0.01
    Negative Logits
     hitter
    -0.64
     nurs
    -0.61
     rehabilit
    -0.60
    abies
    -0.59
     battered
    -0.58
     batter
    -0.57
     veterinarian
    -0.55
    igger
    -0.55
     territ
    -0.54
    estation
    -0.53
    POSITIVE LOGITS
    forth
    2.14
    forward
    1.44
     why
    0.96
    entimes
    0.91
    why
    0.86
    noon
    0.82
    xual
    0.82
    fter
    0.79
    videos
    0.78
    ably
    0.75
    Activations Density 0.027%

    No Known Activations