© 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. 3668
    Prev
    Next
    INDEX
    Explanations

    words related to entities or proper nouns starting with the letter 'T'

    oai_token-act-pair · gpt-3.5-turbo
    Triggered 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
    Neuron Alignment
    Index
    Value
    % of L₁
    659
    +0.16
    0.8%
    167
    +0.10
    0.5%
    116
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    659
    +0.16
    0.07
    213
    +0.10
    0.04
    67
    +0.10
    0.04
    Negative Logits
     ment
    -0.73
    theless
    -0.66
     ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³
    -0.65
     cannabin
    -0.63
     diplom
    -0.62
     destro
    -0.62
    wagen
    -0.61
     intrigue
    -0.59
     descending
    -0.59
     acknow
    -0.58
    POSITIVE LOGITS
    ARGET
    1.32
    ribute
    1.31
    rench
    1.23
    olerance
    1.22
    empt
    1.21
    URN
    1.20
    rav
    1.19
    ributes
    1.18
    akedown
    1.17
    EMP
    1.16
    Activations Density 0.037%

    No Known Activations