© 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. 7945
    Prev
    Next
    INDEX
    Explanations

    social media and online integration-related terms, along with technological features related to social platforms

    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
    132
    +0.10
    0.5%
    603
    +0.10
    0.5%
    113
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    608
    +0.10
    0.06
    113
    +0.10
    0.04
    750
    +0.10
    0.04
    Negative Logits
     Directions
    -0.72
    urban
    -0.67
    vich
    -0.65
    sbm
    -0.64
     WATCHED
    -0.63
    thia
    -0.62
    pard
    -0.61
     Males
    -0.61
     trailing
    -0.59
     followed
    -0.59
    POSITIVE LOGITS
     brink
    1.01
     heights
    1.00
     levels
    0.98
     notch
    0.97
     level
    0.90
     peak
    0.88
     highs
    0.88
     extremes
    0.85
     saturation
    0.85
     threshold
    0.85
    Activations Density 0.296%

    No Known Activations