Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 1536
    5. 8-RES_FS1536-JB
    6. 391
    Prev
    Next
    INDEX
    Explanations

    declarations or requests to provide information

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs1536-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    1,536
    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
    726
    +0.10
    0.5%
    68
    +0.09
    0.4%
    59
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.10
    0.11
    47
    +0.09
    0.06
    1
    +0.08
    0.06
    Negative Logits
    pract
    -0.70
    aji
    -0.70
    peer
    -0.70
    recogn
    -0.67
    rust
    -0.67
    appro
    -0.66
    conduct
    -0.65
    saf
    -0.63
    arma
    -0.62
     interception
    -0.62
    POSITIVE LOGITS
     shelves
    1.02
     vacancies
    1.00
     gaps
    0.98
     coffers
    0.90
     stadiums
    0.86
     arenas
    0.85
     notebooks
    0.84
     slots
    0.83
    leneck
    0.82
     fills
    0.82
    Activations Density 2.437%

    No Known Activations