© 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. 10417
    Prev
    Next
    INDEX
    Explanations

    instances of the word "pour" in the text

    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
    599
    +0.10
    0.5%
    571
    +0.10
    0.5%
    355
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    599
    +0.10
    0.03
    201
    +0.10
    0.03
    245
    +0.10
    0.03
    Negative Logits
    tarians
    -0.78
    nton
    -0.74
    abiding
    -0.68
    obbies
    -0.66
    relation
    -0.65
    ĻĤ
    -0.65
    ALE
    -0.65
    ataka
    -0.64
    locked
    -0.64
    osuke
    -0.64
    POSITIVE LOGITS
     encour
    1.31
     scorn
    0.97
     gallons
    0.93
     forth
    0.92
     boiling
    0.92
     poured
    0.90
     litres
    0.88
     onto
    0.87
     molten
    0.86
     into
    0.84
    Activations Density 0.160%

    No Known Activations