© 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. 20300
    Prev
    Next
    INDEX
    Explanations

    phrases related to heroic actions and rescue operations

    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
    481
    +0.19
    0.9%
    502
    +0.10
    0.5%
    350
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    223
    +0.19
    0.05
    298
    +0.10
    0.04
    121
    +0.10
    0.05
    Negative Logits
     Dating
    -0.71
    Style
    -0.69
     Influ
    -0.69
     Influence
    -0.69
     patents
    -0.68
     Quarterly
    -0.68
     collaborations
    -0.67
     influencing
    -0.65
     influences
    -0.64
     IQ
    -0.63
    POSITIVE LOGITS
     stranded
    1.31
     evacuate
    1.30
     evacuated
    1.26
     evacuation
    1.22
     safely
    1.20
     stret
    1.18
     fleeing
    1.11
     belongings
    1.09
     disembark
    1.09
     evac
    1.07
    Activations Density 0.553%

    No Known Activations