© 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. 10096
    Prev
    Next
    INDEX
    Explanations

    references to airplane accidents or incidents

    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
    741
    +0.13
    0.6%
    346
    +0.13
    0.6%
    484
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    652
    +0.13
    0.02
    346
    +0.13
    0.02
    484
    +0.11
    0.02
    Negative Logits
    FINE
    -0.77
    UGC
    -0.74
    utenberg
    -0.72
    laus
    -0.72
    lishes
    -0.72
    âĶģ
    -0.70
    shire
    -0.69
    PUT
    -0.69
     Democr
    -0.68
    uces
    -0.68
    POSITIVE LOGITS
    walk
    1.12
    walker
    1.11
     wreckage
    1.06
     airliner
    1.06
     crash
    1.06
    wreck
    0.98
     plane
    0.98
     airplanes
    0.97
     prope
    0.97
     flown
    0.97
    Activations Density 0.022%

    No Known Activations