© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW 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. 15603
    Prev
    Next
    INDEX
    Explanations

    mentions of quantity or numbers, especially related to increasing amounts

    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
    132
    +0.15
    0.7%
    741
    +0.09
    0.4%
    402
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    132
    +0.15
    0.03
    402
    +0.09
    0.03
    344
    +0.09
    0.03
    Negative Logits
    odor
    -0.82
    iman
    -0.76
    ISTER
    -0.73
    efe
    -0.70
    ¬¼
    -0.69
    ister
    -0.69
    Gra
    -0.68
    obar
    -0.68
    both
    -0.67
    istry
    -0.67
    POSITIVE LOGITS
     than
    1.16
     installments
    0.91
     layers
    0.86
     rounds
    0.86
     iterations
    0.83
     inches
    0.82
     acres
    0.79
     miles
    0.78
     stringent
    0.77
     arrivals
    0.74
    Activations Density 0.047%

    No Known Activations