© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 10104
    Prev
    Next
    INDEX
    Explanations

    French accents and specialized terms

    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
    +0.24
    1.2%
    447
    +0.18
    0.9%
    406
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    406
    +0.24
    0.07
    659
    +0.18
    0.06
    579
    +0.11
    0.06
    Negative Logits
     severed
    -0.64
     unlaw
    -0.60
     strangers
    -0.60
     coli
    -0.58
     precarious
    -0.58
     Macy
    -0.57
     voluntary
    -0.57
     stub
    -0.57
     horizont
    -0.56
     flowering
    -0.56
    POSITIVE LOGITS
    IUM
    0.90
    anski
    0.87
    coni
    0.84
    heim
    0.83
    Byte
    0.80
    ko
    0.78
    usk
    0.76
    imo
    0.76
    ka
    0.75
    Grade
    0.74
    Activations Density 0.501%

    No Known Activations