© 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. 4196
    Prev
    Next
    INDEX
    Explanations

    phrases related to judgments of extremity and comparison, including terms indicating the most extreme or the most compared to others

    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
    277
    +0.10
    0.5%
    248
    +0.10
    0.5%
    671
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    635
    +0.10
    0.02
    616
    +0.10
    0.02
    256
    +0.10
    0.02
    Negative Logits
    ————
    -0.85
    along
    -0.82
    utra
    -0.76
    berra
    -0.76
    module
    -0.73
    aliation
    -0.70
    arching
    -0.70
    arget
    -0.70
     Retrieved
    -0.68
    ahime
    -0.66
    POSITIVE LOGITS
     slightest
    1.24
     smallest
    1.17
     mundane
    1.02
     curs
    1.01
     simplest
    0.99
     lowly
    0.98
     seasoned
    0.92
     modest
    0.92
     casual
    0.90
     innocuous
    0.89
    Activations Density 0.100%

    No Known Activations