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. 990
    Prev
    Next
    INDEX
    Explanations

    the word "Time" and phrases related to the passing of time

    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
    347
    +0.10
    0.5%
    251
    +0.10
    0.5%
    76
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    651
    +0.10
    0.07
    76
    +0.10
    0.06
    347
    +0.10
    0.06
    Negative Logits
    ourt
    -0.79
    riad
    -0.79
    ramid
    -0.77
    agonist
    -0.69
    warts
    -0.69
    qqa
    -0.69
    eny
    -0.67
    ortium
    -0.67
    iants
    -0.66
    ards
    -0.65
    POSITIVE LOGITS
    frames
    1.27
     elapsed
    1.05
     consuming
    1.01
    zone
    1.00
    frame
    0.99
     lapse
    0.95
     periods
    0.92
    cale
    0.88
    consuming
    0.88
     stamps
    0.87
    Activations Density 0.484%

    No Known Activations