© 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 3072
    9. 8-RES_FS3072-JB
    11. 3019
    Prev
    Next
    INDEX
    Explanations

    statements indicating a controversy or claim

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs3072-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    3,072
    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
    655
    +0.12
    0.6%
    423
    +0.10
    0.5%
    609
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    655
    +0.12
    0.09
    402
    +0.10
    0.08
    609
    +0.09
    0.07
    Negative Logits
     ([
    -0.74
    sburgh
    -0.71
    ocy
    -0.68
     âĵĺ
    -0.67
    :,
    -0.67
    ju
    -0.65
    ellectual
    -0.63
    rive
    -0.62
     ('
    -0.60
    lambda
    -0.58
    POSITIVE LOGITS
     equivalent
    0.88
    lihood
    0.84
    which
    0.79
    nown
    0.78
     unheard
    0.78
     incidentally
    0.77
     testament
    0.76
    arently
    0.74
     euphem
    0.74
     notwithstanding
    0.73
    Activations Density 1.850%

    No Known Activations