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. 511
    Prev
    Next
    INDEX
    Explanations

    financial and medical terms

    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
    479
    +0.15
    0.7%
    481
    +0.14
    0.7%
    496
    +0.13
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    6
    +0.15
    0.14
    640
    +0.14
    0.13
    176
    +0.13
    0.07
    Negative Logits
    rael
    -0.86
    ouston
    -0.79
    ascus
    -0.73
    idine
    -0.72
    hub
    -0.68
    conservancy
    -0.68
    president
    -0.68
     mund
    -0.67
     giants
    -0.66
    aley
    -0.66
    POSITIVE LOGITS
     Contains
    1.26
     Repeat
    1.25
     Perform
    1.20
     Same
    1.19
     Completed
    1.18
     Lack
    1.18
     Repe
    1.15
     Remove
    1.15
     Submit
    1.14
     Advice
    1.14
    Activations Density 1.444%

    No Known Activations