© 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. 1634
    Prev
    Next
    INDEX
    Explanations

    objects or actions related to physical violence

    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
    480
    +0.20
    1.0%
    251
    +0.09
    0.5%
    620
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.20
    0.11
    251
    +0.09
    0.06
    67
    +0.08
    0.07
    Negative Logits
    undrum
    -0.91
    mma
    -0.80
    agame
    -0.76
    eanor
    -0.70
    yssey
    -0.68
    ilogy
    -0.67
    ologue
    -0.65
    swer
    -0.65
    erker
    -0.64
     Originally
    -0.63
    POSITIVE LOGITS
     ones
    0.81
    bugs
    0.73
     emergencies
    0.73
     quantities
    0.69
     groups
    0.69
     bombs
    0.67
     quotas
    0.66
     cafes
    0.66
     alternatives
    0.66
     instruments
    0.66
    Activations Density 0.743%

    No Known Activations