© 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. 5649
    Prev
    Next
    INDEX
    Explanations

    descriptions of technical processes involving specific materials and tools

    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
    424
    +0.10
    0.4%
    520
    +0.09
    0.4%
    575
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.10
    0.08
    575
    +0.09
    0.07
    289
    +0.09
    0.02
    Negative Logits
    bernatorial
    -0.87
    "]=>
    -0.74
    ghazi
    -0.73
    nai
    -0.72
    essage
    -0.66
    emale
    -0.65
     Franchise
    -0.64
    tall
    -0.62
    ropolis
    -0.62
    )=
    -0.61
    POSITIVE LOGITS
     technique
    0.85
     techniques
    0.84
     notation
    0.78
     supplemented
    0.78
     ingenuity
    0.73
     pseudonym
    0.68
     combination
    0.67
     apparatus
    0.67
     algorithm
    0.67
    apult
    0.67
    Activations Density 0.590%

    No Known Activations