© 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 12288
    9. 8-RES_FS12288-JB
    11. 10787
    Prev
    Next
    INDEX
    Explanations

    names of individuals and locations in news articles

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    481
    +0.18
    0.8%
    96
    +0.11
    0.5%
    411
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    96
    +0.18
    0.07
    720
    +0.11
    0.06
    703
    +0.09
    0.04
    Negative Logits
    imum
    -0.81
    natureconservancy
    -0.81
    ongevity
    -0.78
    rastructure
    -0.73
     optimal
    -0.73
    iterranean
    -0.72
    QUIRE
    -0.71
    vana
    -0.71
    arenthood
    -0.70
     optim
    -0.70
    POSITIVE LOGITS
     insults
    1.25
     disrespectful
    1.19
     slurs
    1.19
     abusive
    1.18
     harassing
    1.17
     bullying
    1.16
     tir
    1.15
     insulting
    1.14
     insulted
    1.12
     yelling
    1.11
    Activations Density 1.294%

    No Known Activations