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 24576
    5. 8-RES_FS24576-JB
    6. 23402
    Prev
    Next
    INDEX
    Explanations

    roles and positions of individuals within academic and professional settings

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    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
    373
    +0.12
    0.6%
    599
    +0.12
    0.5%
    741
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    741
    +0.12
    0.04
    579
    +0.12
    0.04
    327
    +0.10
    0.04
    Negative Logits
     brut
    -0.76
     stupidity
    -0.70
    ueller
    -0.69
    rums
    -0.67
     whim
    -0.67
     simplicity
    -0.66
    zees
    -0.66
    humane
    -0.66
     greatness
    -0.66
    ppings
    -0.66
    POSITIVE LOGITS
     adjunct
    1.01
     trustee
    0.99
     Associate
    0.94
     Certified
    0.92
     Coordinator
    0.90
     Editorial
    0.86
    ospons
    0.86
     affiliated
    0.85
     liaison
    0.84
     Consult
    0.83
    Activations Density 0.232%

    No Known Activations