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 3072
    5. 8-RES_FS3072-JB
    6. 1342
    Prev
    Next
    INDEX
    Explanations

    specific mentions of "Part" followed by a numerical value 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_fs3072-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    3,072
    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
    216
    +0.12
    0.5%
    597
    +0.11
    0.5%
    676
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    676
    +0.12
    0.07
    597
    +0.11
    0.06
    216
    +0.11
    0.06
    Negative Logits
    culosis
    -0.69
    berus
    -0.66
     Predators
    -0.59
    stasy
    -0.59
     ceilings
    -0.59
     briefs
    -0.58
     cumbers
    -0.57
    iversal
    -0.56
    incinn
    -0.55
    gently
    -0.55
    POSITIVE LOGITS
    ridge
    1.13
    aking
    0.94
    ridges
    0.94
    ners
    0.93
    meal
    0.92
    icularly
    0.91
    icular
    0.91
    ials
    0.88
    nered
    0.86
    ially
    0.85
    Activations Density 0.390%

    No Known Activations