© 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. 13537
    Prev
    Next
    INDEX
    Explanations

    descriptions of a product, highlighting its features, benefits, and ease of use

    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
    159
    +0.13
    0.6%
    36
    +0.12
    0.5%
    46
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    45
    +0.13
    0.03
    497
    +0.12
    0.04
    159
    +0.11
    0.04
    Negative Logits
     Negro
    -0.70
     Fenrir
    -0.69
     Voices
    -0.69
     Nightmares
    -0.68
    onies
    -0.65
    ortmund
    -0.65
    ority
    -0.64
     fellowship
    -0.64
    izen
    -0.64
    itus
    -0.63
    POSITIVE LOGITS
     inexpensive
    1.29
     efficient
    1.10
     scalable
    1.07
     versatile
    1.06
    usable
    1.02
     economical
    1.02
     reusable
    1.01
     easy
    1.01
     convenient
    1.00
     customizable
    1.00
    Activations Density 0.387%

    No Known Activations