© 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. 8217
    Prev
    Next
    INDEX
    Explanations

    references to time periods and historical data

    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
    664
    +0.11
    0.5%
    349
    +0.11
    0.5%
    306
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    306
    +0.11
    0.05
    651
    +0.11
    0.04
    488
    +0.09
    0.04
    Negative Logits
    umbn
    -0.66
    eny
    -0.58
     twitch
    -0.58
    abase
    -0.56
    boxing
    -0.55
     sympathetic
    -0.54
    yout
    -0.53
    eric
    -0.53
     pissed
    -0.53
    amina
    -0.52
    POSITIVE LOGITS
     lows
    0.71
     onwards
    0.71
     era
    0.70
     eras
    0.69
     epoch
    0.67
     predecessors
    0.65
     preceding
    0.64
    âĢ²
    0.64
    .","
    0.64
    historic
    0.63
    Activations Density 0.201%

    No Known Activations