© 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. 22499
    Prev
    Next
    INDEX
    Explanations

    calls to subscribe or sign up for newsletters or updates

    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
    1
    +0.11
    0.5%
    592
    +0.10
    0.5%
    496
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    592
    +0.11
    0.02
    508
    +0.10
    0.02
    1
    +0.10
    0.01
    Negative Logits
    esan
    -0.90
    inas
    -0.70
     bil
    -0.66
    ball
    -0.65
    come
    -0.65
    stad
    -0.63
    put
    -0.63
    cule
    -0.62
    ilde
    -0.61
    uin
    -0.60
    POSITIVE LOGITS
    CHAT
    0.86
    Subscribe
    0.86
     Subscribe
    0.81
     subscribe
    0.80
    Interstitial
    0.79
     unsub
    0.79
     subscriptions
    0.77
    scribe
    0.77
     RSS
    0.73
    ombat
    0.72
    Activations Density 0.010%

    No Known Activations