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. 23469
    Prev
    Next
    INDEX
    Explanations

    instances of the word "quickly."

    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
    635
    +0.10
    0.4%
    422
    +0.09
    0.4%
    516
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    422
    +0.10
    0.03
    516
    +0.09
    0.02
    635
    +0.08
    0.02
    Negative Logits
    liga
    -0.79
    pel
    -0.78
    anyahu
    -0.71
    ĨĴ
    -0.71
    CHQ
    -0.70
    apixel
    -0.69
    ortmund
    -0.68
    mology
    -0.67
    raints
    -0.67
    ¢
    -0.67
    POSITIVE LOGITS
     paced
    1.01
     enough
    1.00
    ened
    0.91
    idious
    0.88
     approaching
    0.87
    aneously
    0.87
     deterior
    0.83
     thereafter
    0.82
     turnaround
    0.80
     forgotten
    0.79
    Activations Density 0.037%

    No Known Activations