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 3072
    5. 8-RES_FS3072-JB
    6. 1018
    Prev
    Next
    INDEX
    Explanations

    discussion about economic and political issues

    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
    517
    +0.12
    0.6%
    567
    +0.10
    0.4%
    587
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    720
    +0.12
    0.12
    517
    +0.10
    0.10
    73
    +0.09
    0.09
    Negative Logits
    ricane
    -0.74
    issan
    -0.72
    untled
    -0.72
    oward
    -0.71
    aido
    -0.68
    risis
    -0.68
    ricanes
    -0.67
    undai
    -0.66
     newfound
    -0.66
    iously
    -0.65
    POSITIVE LOGITS
     hence
    1.06
     whereas
    1.03
     therefore
    1.01
     insofar
    0.96
     Therefore
    0.92
     meaning
    0.90
     ie
    0.89
     Its
    0.85
     unlike
    0.82
     consisting
    0.81
    Activations Density 2.803%

    No Known Activations