© 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 49152
    9. 8-RES_FS49152-JB
    11. 45704
    Prev
    Next
    INDEX
    Explanations

    statistics and specific numerical performance metrics in sports contexts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    314
    +0.12
    0.5%
    719
    +0.10
    0.5%
    81
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    260
    +0.12
    0.04
    81
    +0.10
    0.04
    198
    +0.10
    0.04
    Negative Logits
     modelling
    -0.65
     Tob
    -0.65
     Rupert
    -0.64
     Wag
    -0.62
    ogyn
    -0.61
     ageing
    -0.60
     Mayer
    -0.60
    staking
    -0.59
     Sinai
    -0.59
    Lago
    -0.58
    POSITIVE LOGITS
    ½
    1.01
    QB
    0.97
    rd
    0.83
    FB
    0.82
    nd
    0.80
     ½
    0.78
    PT
    0.77
    ND
    0.75
    OT
    0.74
    nder
    0.72
    Activations Density 0.071%

    No Known Activations