© 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 12288
    9. 8-RES_FS12288-JB
    11. 10748
    Prev
    Next
    INDEX
    Explanations

    direct or indirect references to the concept of 'thing'

    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
    160
    +0.11
    0.5%
    155
    +0.10
    0.5%
    495
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    737
    +0.11
    0.04
    155
    +0.10
    0.04
    495
    +0.10
    0.03
    Negative Logits
     largeDownload
    -0.68
    ped
    -0.67
    brids
    -0.61
    oufl
    -0.61
    okane
    -0.60
    sts
    -0.60
    claimed
    -0.59
    asers
    -0.59
     turf
    -0.59
     domain
    -0.58
    POSITIVE LOGITS
     happened
    1.15
     happening
    1.13
    iverse
    1.09
     happens
    0.99
     happen
    0.98
     happ
    0.92
     Happ
    0.85
    Else
    0.77
    meal
    0.74
     else
    0.74
    Activations Density 0.078%

    No Known Activations