© 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 49152
    9. 8-RES_FS49152-JB
    11. 41774
    Prev
    Next
    INDEX
    Explanations

    references to ethical and legal violations

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    288
    +0.12
    0.6%
    49
    +0.12
    0.5%
    220
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    49
    +0.12
    0.01
    540
    +0.12
    0.01
    651
    +0.10
    0.01
    Negative Logits
    uana
    -0.86
    zl
    -0.74
    minist
    -0.71
    ICA
    -0.71
    lee
    -0.68
    eele
    -0.68
    arij
    -0.66
    opped
    -0.65
    rose
    -0.64
    onga
    -0.64
    POSITIVE LOGITS
     breaches
    0.96
     breach
    0.91
     confidentiality
    0.78
     breached
    0.77
    hole
    0.76
    hold
    0.76
     Breach
    0.75
    ingly
    0.75
    holes
    0.72
    closure
    0.70
    Activations Density 0.011%

    No Known Activations