© 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. 7045
    Prev
    Next
    INDEX
    Explanations

    terms related to technology like software, SDKs, and features such as speech recognition and development

    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
    447
    +0.16
    0.7%
    679
    +0.10
    0.5%
    225
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.16
    0.05
    575
    +0.10
    0.05
    225
    +0.10
    0.04
    Negative Logits
    birth
    -0.81
    cakes
    -0.79
    ogie
    -0.79
    ewitness
    -0.78
    ruary
    -0.77
    cake
    -0.76
    endon
    -0.76
    zl
    -0.75
     Rebell
    -0.73
    ghazi
    -0.71
    POSITIVE LOGITS
     algorithms
    1.32
     algorithm
    1.17
     capabilities
    1.16
     technology
    1.15
     software
    1.14
     technologies
    1.12
     capability
    1.12
     systems
    1.05
     mechanisms
    1.05
     tools
    1.04
    Activations Density 0.150%

    No Known Activations