Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 1536
    5. 8-RES_FS1536-JB
    6. 527
    Prev
    Next
    INDEX
    Explanations

    time-related phrases, such as durations and points in 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_fs1536-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    1,536
    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
    222
    +0.10
    0.4%
    675
    +0.09
    0.4%
    119
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    737
    +0.10
    0.07
    119
    +0.09
    0.06
    635
    +0.09
    0.06
    Negative Logits
    also
    -0.74
    pora
    -0.72
     Also
    -0.71
    etheus
    -0.69
    successfully
    -0.66
    olon
    -0.66
     likewise
    -0.66
    only
    -0.63
    egal
    -0.62
    achine
    -0.61
    POSITIVE LOGITS
     coincidence
    0.79
    ifiable
    0.76
     superficial
    0.72
     luck
    0.72
     curs
    0.71
     scratched
    0.71
     mortals
    0.70
     penn
    0.70
     cents
    0.69
    hig
    0.68
    Activations Density 2.677%

    No Known Activations