© 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. 24063
    Prev
    Next
    INDEX
    Explanations

    key terms and concepts related to data structures or systems

    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
    694
    +0.11
    0.5%
    710
    +0.10
    0.4%
    2
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    232
    +0.11
    0.07
    288
    +0.10
    0.08
    496
    +0.09
    0.07
    Negative Logits
     vain
    -0.79
    ÄŁ
    -0.68
    eatured
    -0.67
     vigilance
    -0.66
     vest
    -0.66
     pod
    -0.63
     trail
    -0.63
     colle
    -0.62
     shell
    -0.62
     straw
    -0.62
    POSITIVE LOGITS
     Ones
    0.91
     Day
    0.87
     Month
    0.82
     Characters
    0.82
     Manager
    0.81
     Leader
    0.80
     Points
    0.80
     Builder
    0.80
     Definition
    0.79
     Era
    0.79
    Activations Density 0.620%

    No Known Activations