© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW 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. 3963
    Prev
    Next
    INDEX
    Explanations

    phrases related to challenging or difficult circumstances

    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
    411
    +0.11
    0.5%
    453
    +0.11
    0.5%
    730
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    453
    +0.11
    0.02
    411
    +0.11
    0.01
    730
    +0.10
    0.01
    Negative Logits
    assies
    -0.82
    ividual
    -0.81
    ovember
    -0.80
    phis
    -0.74
    iliary
    -0.73
    ĸļ
    -0.73
    ICLE
    -0.72
    alian
    -0.72
    isphere
    -0.71
    cellent
    -0.70
    POSITIVE LOGITS
     winters
    1.05
     punishments
    1.03
    ness
    1.03
     punishment
    1.00
     harsh
    0.97
     retribution
    0.97
    est
    0.94
    nesses
    0.93
     harshly
    0.92
     harsher
    0.88
    Activations Density 0.023%

    No Known Activations