© 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. 1171
    Prev
    Next
    INDEX
    Explanations

    legal and law enforcement terminology

    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
    455
    +0.13
    0.6%
    645
    +0.12
    0.5%
    671
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    221
    +0.13
    0.07
    455
    +0.12
    0.08
    671
    +0.11
    0.06
    Negative Logits
     Conversation
    -0.78
     righteousness
    -0.68
    itaire
    -0.67
     injust
    -0.67
     goodness
    -0.67
     Rebellion
    -0.67
     Colors
    -0.67
    opoly
    -0.65
    Kings
    -0.65
     clich
    -0.65
    POSITIVE LOGITS
     formally
    1.18
     officially
    1.08
     necessarily
    1.00
     disclosed
    0.99
     divul
    0.90
     explicitly
    0.90
     disclose
    0.90
     confir
    0.89
     overtly
    0.89
     immediately
    0.89
    Activations Density 0.327%

    No Known Activations