© 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. 10284
    Prev
    Next
    INDEX
    Explanations

    explanations or definitions beginning with "What is" or "Why is" in technical contexts

    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
    496
    +0.12
    0.6%
    144
    +0.12
    0.5%
    92
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    186
    +0.12
    0.05
    57
    +0.12
    0.04
    60
    +0.10
    0.05
    Negative Logits
     slightest
    -0.67
    ĸļ
    -0.65
    boa
    -0.62
    general
    -0.62
    fm
    -0.61
    Ire
    -0.61
    gi
    -0.60
    ONSORED
    -0.60
    efully
    -0.60
    ofi
    -0.59
    POSITIVE LOGITS
     Makes
    0.92
     Difference
    0.91
     Does
    0.90
     Definitions
    0.88
     Stuff
    0.84
     Facts
    0.83
     Picks
    0.82
     Benefits
    0.82
     Documents
    0.80
     Exactly
    0.79
    Activations Density 0.242%

    No Known Activations