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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 22614
    Prev
    Next
    INDEX
    Explanations

    references to "blue" in various contexts

    oai_token-act-pair · gpt-4o-miniTriggered 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
    70
    +0.10
    0.4%
    12
    +0.10
    0.4%
    14
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    12
    +0.10
    0.02
    188
    +0.10
    0.02
    525
    +0.10
    0.02
    Negative Logits
    ROR
    -0.78
    rolet
    -0.77
    NRS
    -0.76
    IFA
    -0.72
    Ö¼
    -0.71
    ETF
    -0.70
    abulary
    -0.70
    uador
    -0.69
    ernandez
    -0.69
    ITNESS
    -0.69
    POSITIVE LOGITS
    prints
    1.47
    grass
    1.31
    berry
    1.27
    berries
    1.14
    beard
    1.08
    bird
    1.06
     ribbon
    1.01
    fin
    0.96
     dye
    0.96
     eyed
    0.94
    Activations Density 0.015%

    No Known Activations