© 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. 8597
    Prev
    Next
    INDEX
    Explanations

    mentions of criticism or conflict

    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
    296
    +0.12
    0.5%
    720
    +0.12
    0.5%
    688
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    720
    +0.12
    0.06
    472
    +0.12
    0.04
    69
    +0.10
    0.04
    Negative Logits
     ILCS
    -0.79
     husbands
    -0.73
    husband
    -0.66
    )].
    -0.64
    arers
    -0.64
    ante
    -0.64
    selves
    -0.61
     metre
    -0.61
     GOODMAN
    -0.61
    ahi
    -0.60
    POSITIVE LOGITS
     pard
    0.89
     pardon
    0.80
     Mexicans
    0.80
     Vladimir
    0.78
     Russia
    0.78
     his
    0.77
     Pence
    0.75
     Putin
    0.75
     himself
    0.74
     Monica
    0.74
    Activations Density 0.580%

    No Known Activations