© 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 24576
    9. 8-RES_FS24576-JB
    11. 11969
    Prev
    Next
    INDEX
    Explanations

    references to different types of media such as TV shows, movies, books, and anime

    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
    447
    +0.18
    0.9%
    459
    +0.12
    0.5%
    481
    +0.12
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    563
    +0.18
    0.07
    217
    +0.12
    0.06
    455
    +0.12
    0.07
    Negative Logits
    dden
    -0.67
     Helpful
    -0.59
     forgetting
    -0.59
    sbm
    -0.56
    Unknown
    -0.56
    Redd
    -0.55
     Wrong
    -0.55
    ãģ¦
    -0.54
    dropping
    -0.54
     understatement
    -0.54
    POSITIVE LOGITS
     consists
    1.29
     consisted
    1.28
     comprises
    1.16
     revolves
    1.15
     debuted
    1.15
     contains
    1.03
     boasts
    1.02
     premiered
    1.02
     underwent
    1.01
     originated
    1.01
    Activations Density 0.324%

    No Known Activations