© 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 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. 17612
    Prev
    Next
    INDEX
    Explanations

    news-related text, specifically focusing on reports and updates

    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
    640
    +0.28
    1.3%
    87
    +0.12
    0.6%
    120
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    640
    +0.28
    0.14
    94
    +0.12
    0.11
    64
    +0.10
    -0.04
    Negative Logits
     footing
    -0.69
    lections
    -0.68
     illusion
    -0.65
    burn
    -0.65
    ensibly
    -0.64
    minist
    -0.63
     uninterrupted
    -0.62
     lifes
    -0.61
     manoeuv
    -0.60
     inadequate
    -0.60
    POSITIVE LOGITS
    SOURCE
    0.88
    Featured
    0.85
    Meanwhile
    0.84
    <|endoftext|>
    0.84
    =-=-=-=-=-=-=-=-
    0.83
    PHOTOS
    0.82
    Stay
    0.82
    Also
    0.81
    ©
    0.80
    Related
    0.80
    Activations Density 0.396%

    No Known Activations