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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & 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. 26407
    Prev
    Next
    INDEX
    Explanations

    the presence of the word "with" and its various contexts

    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
    197
    +0.10
    0.5%
    640
    +0.10
    0.5%
    362
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    60
    +0.10
    0.04
    362
    +0.10
    0.04
    498
    +0.09
    0.04
    Negative Logits
    thin
    -0.84
    sburg
    -0.80
    station
    -0.72
     Awakens
    -0.68
    fair
    -0.68
    deen
    -0.66
    book
    -0.66
    hops
    -0.66
    TG
    -0.65
    fuck
    -0.64
    POSITIVE LOGITS
     regard
    1.11
    drawn
    1.09
    stood
    1.07
     regards
    0.96
    draw
    0.94
     assistance
    0.93
     respect
    0.86
     hopes
    0.85
     impunity
    0.85
     apologies
    0.84
    Activations Density 0.117%

    No Known Activations