© 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. 17794
    Prev
    Next
    INDEX
    Explanations

    phrases related to legal or political discussions, specifically around granting status and organizations

    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
    16
    +0.11
    0.5%
    456
    +0.11
    0.5%
    218
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    16
    +0.11
    0.01
    588
    +0.11
    0.01
    232
    +0.10
    0.01
    Negative Logits
    SPONSORED
    -0.72
     4090
    -0.68
    renheit
    -0.63
    oteric
    -0.62
    Story
    -0.61
    OTOS
    -0.60
     Hert
    -0.60
     Bei
    -0.60
     Xan
    -0.59
     fundament
    -0.59
    POSITIVE LOGITS
    uced
    1.33
    ding
    1.27
    eem
    1.27
    ucing
    1.21
    irect
    1.13
    uce
    1.12
    uction
    1.11
    ded
    1.10
    icative
    1.10
    acted
    1.09
    Activations Density 0.021%

    No Known Activations