© 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. 14119
    Prev
    Next
    INDEX
    Explanations

    words related to geographical locations

    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
    1
    +0.13
    0.6%
    616
    +0.10
    0.5%
    214
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    4
    +0.13
    0.04
    616
    +0.10
    0.03
    767
    +0.10
    0.03
    Negative Logits
     Cruiser
    -0.65
     Jiu
    -0.60
     Kerr
    -0.56
     citation
    -0.55
    WOOD
    -0.55
     Crusade
    -0.55
    Bey
    -0.54
     Portuguese
    -0.54
    Reviewer
    -0.53
     nick
    -0.53
    POSITIVE LOGITS
    cript
    1.31
    ources
    1.23
    atisf
    1.21
    cence
    1.20
    hip
    1.19
    peed
    1.19
    ilver
    1.19
    chool
    1.19
    pect
    1.18
    ystem
    1.15
    Activations Density 0.102%

    No Known Activations