© 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. 22853
    Prev
    Next
    INDEX
    Explanations

    references to security alerts and notifications regarding potential threats

    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
    491
    +0.13
    0.6%
    178
    +0.10
    0.4%
    196
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    703
    +0.13
    0.03
    491
    +0.10
    0.03
    458
    +0.10
    0.03
    Negative Logits
     uncomp
    -0.73
     foundations
    -0.69
     heap
    -0.68
     enrich
    -0.67
     richest
    -0.65
     tort
    -0.64
     collaborations
    -0.64
     impover
    -0.64
    utenberg
    -0.63
    Bench
    -0.63
    POSITIVE LOGITS
     alerts
    1.28
     alarms
    1.24
     alert
    1.16
     alerted
    1.13
    Alert
    1.11
     relayed
    1.11
     dispatcher
    1.06
     ringing
    1.06
     warning
    1.05
     beacon
    1.04
    Activations Density 0.526%

    No Known Activations