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 24576
    5. 8-RES_FS24576-JB
    6. 659
    Prev
    Next
    INDEX
    Explanations

    articles discussing warnings or recommendations, particularly related to health and safety

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    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
    373
    +0.14
    0.7%
    408
    +0.10
    0.4%
    660
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    660
    +0.14
    0.02
    337
    +0.10
    0.02
    650
    +0.09
    0.02
    Negative Logits
    oris
    -0.81
    vier
    -0.75
    nor
    -0.71
    cloth
    -0.69
    sav
    -0.69
    contract
    -0.69
    cos
    -0.68
    iliary
    -0.68
    existent
    -0.67
    orate
    -0.67
    POSITIVE LOGITS
     raining
    0.82
     roaring
    0.72
     flooding
    0.68
     closer
    0.67
     undone
    0.67
     pouring
    0.67
     tempting
    0.64
     downhill
    0.64
     comparing
    0.63
     temptation
    0.60
    Activations Density 0.037%

    No Known Activations