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 768
    5. 8-RES_FS768-JB
    6. 610
    Prev
    Next
    INDEX
    Explanations

    political and governmental terms and topics

    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
    616
    +0.12
    0.5%
    672
    +0.10
    0.5%
    7
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    672
    +0.12
    0.07
    79
    +0.10
    0.12
    462
    +0.10
    0.09
    Negative Logits
     suggestion
    -0.86
     Squadron
    -0.79
     suggestions
    -0.75
     bribes
    -0.67
     Removal
    -0.67
     pedals
    -0.67
     advice
    -0.65
    PLIED
    -0.64
     assistants
    -0.64
    Action
    -0.64
    POSITIVE LOGITS
    population
    1.02
     populous
    0.96
     borders
    0.95
     ravaged
    0.95
     stagn
    0.94
     industrialized
    0.91
     invaded
    0.91
     economically
    0.91
     prosperous
    0.90
     populated
    0.90
    Activations Density 7.601%

    No Known Activations