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 49152
    5. 8-RES_FS49152-JB
    6. 37077
    Prev
    Next
    INDEX
    Explanations

    names of individuals and associated activities in contexts involving human rights

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    314
    +0.13
    0.6%
    659
    +0.11
    0.5%
    728
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    659
    +0.13
    0.06
    232
    +0.11
    0.04
    284
    +0.11
    0.04
    Negative Logits
     NX
    -0.88
     Californ
    -0.83
     FANTASY
    -0.82
     Titanic
    -0.82
     Uncharted
    -0.82
     Interstellar
    -0.81
     conventions
    -0.76
     ponies
    -0.76
     revolving
    -0.75
     Jurassic
    -0.74
    POSITIVE LOGITS
    llah
    1.33
    aq
    1.29
     Hasan
    1.27
    azi
    1.23
    ullah
    1.21
    jad
    1.20
    hammad
    1.20
    rahim
    1.16
     Muhammad
    1.14
     Hassan
    1.13
    Activations Density 0.092%

    No Known Activations