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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 24445
    Prev
    Next
    INDEX
    Explanations

    phrases that express a qualitative evaluation or opinion, particularly regarding experiences or ideas

    oai_token-act-pair · gpt-4o-miniTriggered 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
    377
    +0.12
    0.5%
    393
    +0.11
    0.5%
    362
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    36
    +0.12
    0.04
    362
    +0.11
    0.03
    512
    +0.10
    0.02
    Negative Logits
    milo
    -0.92
    "]=>
    -0.74
    ounty
    -0.74
    anooga
    -0.72
     ashore
    -0.72
    */(
    -0.70
    bral
    -0.70
     eleph
    -0.68
    ARS
    -0.68
    soever
    -0.68
    POSITIVE LOGITS
     contradiction
    1.05
     kindred
    0.85
     distant
    0.81
     quaint
    0.80
     innocuous
    0.80
     oxy
    0.80
     coincidence
    0.79
     dystopian
    0.77
     mismatch
    0.77
     inev
    0.76
    Activations Density 0.095%

    No Known Activations