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 1536
    5. 8-RES_FS1536-JB
    6. 363
    Prev
    Next
    INDEX
    Explanations

    technical terms and corporate concepts in a business or industry context

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs1536-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    1,536
    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
    480
    +0.50
    2.6%
    447
    +0.12
    0.6%
    266
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    480
    +0.50
    0.01
    266
    +0.12
    0.26
    351
    +0.12
    0.06
    Negative Logits
    hovah
    -0.71
    currently
    -0.61
     varies
    -0.60
    erenn
    -0.59
    eware
    -0.59
     awaits
    -0.58
     currently
    -0.58
     WATCH
    -0.58
     depends
    -0.58
    ctuary
    -0.56
    POSITIVE LOGITS
    didn
    0.78
    Was
    0.76
     hadn
    0.74
     consisted
    0.72
     Was
    0.71
     lacked
    0.70
    was
    0.69
     knew
    0.68
     rushed
    0.68
    Had
    0.68
    Activations Density 9.281%

    No Known Activations