© 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. 5117
    Prev
    Next
    INDEX
    Explanations

    phrases related to government agencies and management

    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
    480
    +0.67
    4.1%
    87
    +0.16
    1.0%
    266
    +0.14
    0.9%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    480
    +0.67
    0.03
    266
    +0.16
    0.08
    232
    +0.14
    0.07
    Negative Logits
     UNCLASSIFIED
    -0.49
     interchange
    -0.46
     corro
    -0.46
    enegger
    -0.43
    rawdownloadcloneembedreportprint
    -0.42
     slang
    -0.39
     jargon
    -0.39
     Schn
    -0.39
     decom
    -0.39
     evid
    -0.39
    POSITIVE LOGITS
    çİĭ
    0.63
    ©
    0.60
    ®
    0.58
    º
    0.55
    ·
    0.55
    ÙĬ
    0.54
    »
    0.54
    £
    0.52
    IJ
    0.52
    ãĤ«
    0.51
    Activations Density 0.813%

    No Known Activations