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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlogSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 3072
    5. 8-RES_FS3072-JB
    6. 1704
    Prev
    Next
    INDEX
    Explanations

    financial data and statistics

    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
    369
    +0.10
    0.5%
    43
    +0.10
    0.5%
    297
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    43
    +0.10
    0.14
    546
    +0.10
    0.15
    765
    +0.10
    0.14
    Negative Logits
    reated
    -0.80
    rio
    -0.75
    izen
    -0.72
    ire
    -0.71
    sein
    -0.70
    ¥
    -0.70
    inar
    -0.68
    ¬¼
    -0.67
    irie
    -0.65
    ong
    -0.65
    POSITIVE LOGITS
     whereas
    1.50
     although
    1.34
     but
    1.22
     meaning
    1.11
     prompting
    1.10
     indicating
    1.10
     respectively
    1.09
     implying
    1.09
     though
    1.08
     albeit
    1.07
    Activations Density 1.492%

    No Known Activations