© 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. 2140
    Prev
    Next
    INDEX
    Explanations

    references to specific clothing items

    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
    373
    +0.14
    0.7%
    26
    +0.12
    0.6%
    168
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    26
    +0.14
    0.01
    659
    +0.12
    0.01
    701
    +0.10
    0.01
    Negative Logits
     Declaration
    -0.70
     stakes
    -0.61
     vigil
    -0.61
     superficial
    -0.59
     anx
    -0.59
     Dame
    -0.59
     hazards
    -0.59
     Antiqu
    -0.58
     Hes
    -0.58
     theology
    -0.57
    POSITIVE LOGITS
    ernaut
    1.13
    arette
    1.10
    iants
    1.03
    hetto
    1.03
    regor
    1.00
    raph
    0.99
    asus
    0.98
    hetti
    0.95
    raft
    0.94
    raphics
    0.94
    Activations Density 0.021%

    No Known Activations