© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    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. 10046
    Prev
    Next
    INDEX
    Explanations

    references to injuries or wounds

    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
    375
    +0.13
    0.6%
    211
    +0.11
    0.5%
    420
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    211
    +0.13
    0.01
    375
    +0.11
    0.01
    315
    +0.11
    0.01
    Negative Logits
    ullivan
    -0.74
    essions
    -0.71
    anooga
    -0.70
     GEAR
    -0.70
    ructose
    -0.69
     deviations
    -0.68
    authorized
    -0.67
    ramid
    -0.65
    Sche
    -0.65
    alist
    -0.64
    POSITIVE LOGITS
     wounds
    1.38
     inflicted
    1.26
     wound
    1.15
     healed
    1.01
     healing
    0.96
     scars
    0.94
     heals
    0.92
    bite
    0.91
     heal
    0.90
     wounded
    0.83
    Activations Density 0.030%

    No Known Activations