© 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. 459
    Prev
    Next
    INDEX
    Explanations

    comparisons and contrasts between different things or concepts

    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
    355
    +0.12
    0.5%
    22
    +0.09
    0.4%
    16
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    355
    +0.12
    0.07
    362
    +0.09
    0.07
    408
    +0.09
    0.07
    Negative Logits
    orsi
    -0.85
    orer
    -0.73
    ahi
    -0.68
    orem
    -0.65
    gey
    -0.63
    utherford
    -0.62
    apego
    -0.62
    illard
    -0.62
     awoken
    -0.62
    wives
    -0.61
    POSITIVE LOGITS
     manner
    1.43
     fashion
    1.23
     vein
    1.18
     guise
    1.16
     increments
    1.15
     contexts
    1.11
     context
    1.11
     vicinity
    1.05
     circumstances
    1.02
     hurry
    1.00
    Activations Density 0.436%

    No Known Activations