© 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 768
    9. 8-RES_FS768-JB
    11. 729
    Prev
    Next
    INDEX
    Explanations

    phrases related to specific events or actions starting with the word "When"

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs768-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    768
    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
    476
    +0.11
    0.5%
    706
    +0.11
    0.5%
    98
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    582
    +0.11
    0.08
    476
    +0.11
    0.08
    98
    +0.10
    0.10
    Negative Logits
    agin
    -0.79
    zzi
    -0.75
    bear
    -0.66
    whatever
    -0.65
    kaya
    -0.65
    aido
    -0.65
    hack
    -0.63
    stead
    -0.63
    ictive
    -0.62
    aking
    -0.62
    POSITIVE LOGITS
    soever
    1.36
     confronted
    1.09
     faced
    0.98
     asked
    0.95
     pressed
    0.95
     encountering
    0.87
     evaluating
    0.81
     comparing
    0.81
     viewed
    0.81
     calculating
    0.80
    Activations Density 1.541%

    No Known Activations