© 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. 9743
    Prev
    Next
    INDEX
    Explanations

    references to specific events and topics, such as political controversies or historical contexts in various documents

    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
    87
    +0.26
    1.3%
    424
    +0.10
    0.5%
    127
    +0.09
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    362
    +0.26
    0.08
    608
    +0.10
    0.07
    385
    +0.09
    0.06
    Negative Logits
    pmwiki
    -0.66
    arse
    -0.59
    ciplinary
    -0.59
    wcsstore
    -0.56
    ibo
    -0.56
    entimes
    -0.56
     besides
    -0.55
     sidx
    -0.55
     individually
    -0.53
    zens
    -0.53
    POSITIVE LOGITS
     steroids
    0.74
     Cambodia
    0.64
     Venice
    0.64
     Uganda
    0.63
     Constantinople
    0.63
     Camel
    0.62
     Yugoslavia
    0.60
     which
    0.60
     America
    0.59
    rium
    0.58
    Activations Density 0.571%

    No Known Activations