Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlogSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 49152
    5. 8-RES_FS49152-JB
    6. 37098
    Prev
    Next
    INDEX
    Explanations

    instances of the word "cup" and its variations

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    35
    +0.12
    0.5%
    715
    +0.12
    0.5%
    526
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    35
    +0.12
    0.01
    715
    +0.12
    0.01
    19
    +0.11
    0.01
    Negative Logits
    eering
    -0.76
     Giuliani
    -0.71
     Voc
    -0.67
     partName
    -0.63
    ccording
    -0.63
    IGHTS
    -0.62
     gearing
    -0.62
     reluct
    -0.62
     generational
    -0.61
    SPONSORED
    -0.61
    POSITIVE LOGITS
    cake
    1.94
    cakes
    1.86
    cup
    1.14
    board
    1.09
    boards
    1.05
    idity
    0.99
    illo
    0.91
    ener
    0.90
     cups
    0.90
    atu
    0.89
    Activations Density 0.008%

    No Known Activations