© 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 EvalsExportsinterp-engineNEWAPI 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. 22354
    Prev
    Next
    INDEX
    Explanations

    terms related to gaming beta testing and release processes

    oai_token-act-pair · gpt-4o-miniTriggered 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
    660
    +0.12
    0.5%
    447
    +0.12
    0.5%
    651
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    651
    +0.12
    0.05
    594
    +0.12
    0.04
    288
    +0.11
    0.05
    Negative Logits
    chi
    -0.77
    lasses
    -0.70
    column
    -0.70
    relative
    -0.69
    isine
    -0.68
    uku
    -0.66
    abul
    -0.65
    equal
    -0.65
    si
    -0.64
     unfocusedRange
    -0.62
    POSITIVE LOGITS
     prototypes
    1.19
     testers
    1.17
     rollout
    1.17
     prototype
    1.13
     testing
    1.13
     phase
    1.03
    prototype
    1.00
     demo
    0.99
     preview
    0.95
     trials
    0.93
    Activations Density 0.253%

    No Known Activations