© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 24576
    9. 8-RES_FS24576-JB
    11. 8559
    Prev
    Next
    INDEX
    Explanations

    references to contrasting options or aspects within a scenario

    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
    393
    +0.10
    0.5%
    485
    +0.10
    0.5%
    271
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    469
    +0.10
    0.02
    621
    +0.10
    0.02
    271
    +0.09
    0.02
    Negative Logits
     1915
    -0.72
    udence
    -0.59
     1903
    -0.57
     1919
    -0.56
     1906
    -0.55
     1918
    -0.55
     reintrodu
    -0.55
     1961
    -0.55
     2024
    -0.54
     1912
    -0.54
    POSITIVE LOGITS
    worldly
    1.90
     than
    1.09
    wise
    1.07
    etheless
    0.95
    ials
    0.89
    Redd
    0.88
    than
    0.87
    ially
    0.85
    parts
    0.83
     Than
    0.80
    Activations Density 0.049%

    No Known Activations