© 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 EvalsExportsAPI 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. 12553
    Prev
    Next
    INDEX
    Explanations

    words related to controversy or conflict

    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
    634
    +0.10
    0.5%
    98
    +0.10
    0.5%
    404
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    490
    +0.10
    0.05
    98
    +0.10
    0.05
    438
    +0.09
    0.04
    Negative Logits
    yne
    -0.74
    çīĪ
    -0.70
    artments
    -0.63
    ļéĨĴ
    -0.63
    fif
    -0.63
     flourished
    -0.62
    ²¾
    -0.60
    ngth
    -0.58
    ircraft
    -0.58
    axies
    -0.58
    POSITIVE LOGITS
     ado
    0.82
     anymore
    0.82
     GamerGate
    0.76
     trend
    0.66
    izable
    0.66
    NAS
    0.65
     negativity
    0.64
     deterrent
    0.64
    acea
    0.64
     trolling
    0.64
    Activations Density 0.600%

    No Known Activations