© 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. 26659
    Prev
    Next
    INDEX
    Explanations

    terms related to predictive, descriptive, and diagnostic qualities in data or discussions

    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
    713
    +0.13
    0.6%
    362
    +0.12
    0.5%
    453
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.13
    0.03
    711
    +0.12
    0.02
    713
    +0.11
    0.02
    Negative Logits
    eem
    -0.98
    ulhu
    -0.94
    cedented
    -0.89
    olen
    -0.86
    buck
    -0.85
    ocrates
    -0.83
    olina
    -0.83
    elo
    -0.82
    puter
    -0.82
    igon
    -0.79
    POSITIVE LOGITS
     qualities
    0.98
     impulse
    0.96
     behaviour
    0.95
     impulses
    0.94
     measures
    0.93
     activity
    0.91
     behavior
    0.91
     intent
    0.90
     tendencies
    0.88
     minded
    0.87
    Activations Density 0.057%

    No Known Activations