© 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. 1208
    Prev
    Next
    INDEX
    Explanations

    phrases related to time frames or time periods

    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
    373
    +0.11
    0.5%
    676
    +0.10
    0.4%
    284
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    676
    +0.11
    0.03
    312
    +0.10
    0.03
    651
    +0.10
    0.03
    Negative Logits
     Franch
    -0.75
    wagon
    -0.71
    atism
    -0.71
    BuyableInstoreAndOnline
    -0.71
    hat
    -0.70
    starting
    -0.69
    NEY
    -0.66
    hyde
    -0.66
    rolling
    -0.66
    racuse
    -0.65
    POSITIVE LOGITS
    ebin
    1.51
     tense
    1.24
     decade
    1.20
     few
    0.95
     century
    0.92
     week
    0.92
     fortnight
    0.92
    orate
    0.90
     month
    0.90
     millennium
    0.88
    Activations Density 0.036%

    No Known Activations