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. 1413
    Prev
    Next
    INDEX
    Explanations

    dates in the format month year

    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
    87
    +0.54
    4.3%
    464
    +0.23
    1.8%
    756
    +0.23
    1.8%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    87
    +0.54
    -0.04
    756
    +0.23
    -0.07
    459
    +0.23
    0.26
    Negative Logits
    DragonMagazine
    -0.44
    ¶
    -0.40
    RESULTS
    -0.37
    âķIJâķIJ
    -0.34
     Canaver
    -0.33
    PsyNetMessage
    -0.30
    ActionCode
    -0.30
    -+-+
    -0.30
    ü
    -0.30
    õ
    -0.30
    POSITIVE LOGITS
    heid
    0.35
     thereof
    0.34
    ibur
    0.33
     entit
    0.32
     displ
    0.31
    milo
    0.30
     choke
    0.29
     insure
    0.29
     latter
    0.28
    nostic
    0.28
    Activations Density 17.328%

    No Known Activations