© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    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. 5673
    Prev
    Next
    INDEX
    Explanations

    JavaScript functions

    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
    213
    +0.10
    0.5%
    211
    +0.10
    0.5%
    392
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    211
    +0.10
    0.01
    712
    +0.10
    0.02
    540
    +0.09
    0.02
    Negative Logits
    cott
    -0.84
     Lank
    -0.71
    creen
    -0.70
     Hemp
    -0.67
     Sons
    -0.66
     Signs
    -0.65
    jen
    -0.65
     Jed
    -0.64
    pee
    -0.64
     Lyn
    -0.64
    POSITIVE LOGITS
    ality
    1.03
    alities
    0.92
    aries
    0.92
    ional
    0.91
    ivity
    0.90
    ative
    0.85
    ioned
    0.85
    inic
    0.84
    istry
    0.81
    nel
    0.79
    Activations Density 0.020%

    No Known Activations