© 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 768
    9. 8-RES_FS768-JB
    11. 705
    Prev
    Next
    INDEX
    Explanations

    references to writing activities such as writing books, journal articles, and letters

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs768-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    768
    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
    567
    +0.10
    0.5%
    741
    +0.10
    0.5%
    246
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    76
    +0.10
    0.12
    346
    +0.10
    0.11
    651
    +0.10
    0.11
    Negative Logits
    Ĭ±
    -0.80
    agara
    -0.76
    xon
    -0.75
    nel
    -0.71
    nels
    -0.70
    asia
    -0.69
    rises
    -0.69
    HUD
    -0.68
    roo
    -0.68
    avior
    -0.67
    POSITIVE LOGITS
    lishing
    1.12
     manuscript
    1.12
    itatively
    1.10
     aloud
    1.10
     excerpts
    1.03
     books
    1.01
     essays
    0.98
     penned
    0.96
    writing
    0.96
     manuscripts
    0.95
    Activations Density 4.600%

    No Known Activations