© 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. 1807
    Prev
    Next
    INDEX
    Explanations

    technical terms related to data manipulation and analysis, such as 'persisted data', 'compose files', and 'docker-compose'

    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
    87
    +0.24
    1.2%
    756
    +0.17
    0.9%
    330
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    756
    +0.24
    -0.01
    87
    +0.17
    -0.04
    283
    +0.10
    0.01
    Negative Logits
    \\\\\\\\
    -0.74
    atform
    -0.63
    issance
    -0.63
    ãĥĦ
    -0.57
    EH
    -0.56
     itself
    -0.56
    NES
    -0.56
    ODUCT
    -0.56
    Actor
    -0.56
    Ö¼
    -0.55
    POSITIVE LOGITS
     abound
    0.97
     prolifer
    0.88
     clustered
    0.78
    were
    0.75
     converge
    0.74
     spaced
    0.71
     plentiful
    0.68
     expire
    0.65
     circulate
    0.64
     piled
    0.64
    Activations Density 1.886%

    No Known Activations