© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 24576
    9. 8-RES_FS24576-JB
    11. 18937
    Prev
    Next
    INDEX
    Explanations

    phrases related to worsening or getting better over 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_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
    665
    +0.12
    0.6%
    202
    +0.10
    0.5%
    282
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    272
    +0.12
    0.03
    119
    +0.10
    0.03
    665
    +0.10
    0.03
    Negative Logits
    ij士
    -0.68
     entirety
    -0.66
    Zone
    -0.63
    ellow
    -0.62
    xtap
    -0.61
    ouch
    -0.61
    nav
    -0.60
    efe
    -0.59
    ¬¼
    -0.59
    invoke
    -0.59
    POSITIVE LOGITS
     progressively
    1.00
     exponentially
    0.89
     veter
    0.87
     retard
    0.81
     gradually
    0.79
     sooner
    0.79
     steadily
    0.78
     faster
    0.77
     tires
    0.77
     quicker
    0.76
    Activations Density 0.088%

    No Known Activations