© 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. 22680
    Prev
    Next
    INDEX
    Explanations

    references to military threats and conflicts

    oai_token-act-pair · gpt-4o-miniTriggered 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
    756
    +0.10
    0.5%
    203
    +0.10
    0.5%
    218
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    298
    +0.10
    0.05
    601
    +0.10
    0.03
    608
    +0.10
    0.07
    Negative Logits
     Typ
    -0.78
     anecd
    -0.76
    notation
    -0.74
     Libraries
    -0.73
     Loft
    -0.72
     Cosponsors
    -0.71
     Studio
    -0.71
    College
    -0.70
    Studio
    -0.70
     annotations
    -0.69
    POSITIVE LOGITS
     terrorist
    1.25
     destabil
    1.24
     terrorists
    1.23
     insurgency
    1.21
     overthrow
    1.21
     jihad
    1.20
     milit
    1.19
     Jihad
    1.18
     terrorism
    1.18
     bloodshed
    1.18
    Activations Density 0.783%

    No Known Activations