© 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. 10477
    Prev
    Next
    INDEX
    Explanations

    positive emotional responses and expressions of satisfaction

    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
    459
    +0.11
    0.5%
    496
    +0.09
    0.4%
    610
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    317
    +0.11
    0.04
    242
    +0.09
    0.04
    259
    +0.09
    0.03
    Negative Logits
    perse
    -0.80
    pering
    -0.71
    mage
    -0.66
    uli
    -0.65
    adish
    -0.65
    ngth
    -0.65
    effic
    -0.64
    rawl
    -0.64
    ciplinary
    -0.63
    cipline
    -0.63
    POSITIVE LOGITS
     enough
    0.69
    dy
    0.68
     Wynne
    0.66
     seeing
    0.65
     they
    0.64
     surprises
    0.64
     congr
    0.63
     tid
    0.62
    fully
    0.61
     Ichigo
    0.61
    Activations Density 0.136%

    No Known Activations