© 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. 9609
    Prev
    Next
    INDEX
    Explanations

    phrases related to tracking tasks or activities over time

    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
    60
    +0.14
    0.7%
    648
    +0.10
    0.4%
    16
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    60
    +0.14
    0.05
    654
    +0.10
    0.04
    686
    +0.09
    0.03
    Negative Logits
    krit
    -0.70
     sadly
    -0.69
    nesty
    -0.67
    Politics
    -0.64
    knit
    -0.62
    albeit
    -0.61
     Canaver
    -0.60
    uba
    -0.60
    political
    -0.60
     furious
    -0.60
    POSITIVE LOGITS
     nearest
    0.94
     specified
    0.93
     selected
    0.90
     desired
    0.90
     respective
    0.82
     each
    0.82
    icipated
    0.80
     corresponds
    0.79
     corresponding
    0.77
     Selected
    0.77
    Activations Density 0.483%

    No Known Activations