© 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. 22137
    Prev
    Next
    INDEX
    Explanations

    terms related to flexibility and organizational efficiency in a system

    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
    87
    +0.23
    1.1%
    51
    +0.13
    0.6%
    450
    +0.09
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    51
    +0.23
    0.06
    756
    +0.13
    -0.02
    513
    +0.09
    0.04
    Negative Logits
    âĢİ
    -0.60
     Latest
    -0.56
    roy
    -0.55
     Bars
    -0.54
    wings
    -0.54
     awoke
    -0.54
     (£
    -0.54
    president
    -0.53
     âĢİ
    -0.53
    Newsletter
    -0.52
    POSITIVE LOGITS
     behavi
    0.73
     overall
    0.73
     usability
    0.72
     tremend
    0.72
     valuable
    0.71
    effic
    0.70
     incentive
    0.70
     hemor
    0.70
     unwanted
    0.68
     debugging
    0.67
    Activations Density 0.785%

    No Known Activations