© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 12288
    9. 8-RES_FS12288-JB
    11. 10165
    Prev
    Next
    INDEX
    Explanations

    statistics and numbers in news articles

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    512
    +0.12
    0.6%
    364
    +0.11
    0.5%
    314
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    153
    +0.12
    0.05
    512
    +0.11
    0.05
    488
    +0.10
    0.05
    Negative Logits
     Gorge
    -0.65
     alike
    -0.59
     pant
    -0.57
     retina
    -0.56
     GF
    -0.53
     Minor
    -0.53
     pat
    -0.51
     VERY
    -0.51
     whim
    -0.51
     soul
    -0.50
    POSITIVE LOGITS
     previous
    0.82
     earlier
    0.82
    vous
    0.80
    erest
    0.77
     previously
    0.75
    velt
    0.75
    erva
    0.74
     predecessors
    0.72
     mosqu
    0.71
     predec
    0.71
    Activations Density 0.364%

    No Known Activations