© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW 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. 2985
    Prev
    Next
    INDEX
    Explanations

    phrases indicating a negative or strongly oppositional stance towards a specific subject or entity

    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
    373
    +0.11
    0.5%
    572
    +0.10
    0.5%
    348
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    348
    +0.11
    0.02
    572
    +0.10
    0.02
    512
    +0.10
    0.02
    Negative Logits
    lance
    -0.75
    plex
    -0.70
    riber
    -0.67
    kamp
    -0.66
    (){
    -0.66
    ãĤ¤ãĥĪ
    -0.65
    itor
    -0.65
    place
    -0.65
    iery
    -0.65
    plates
    -0.64
    POSITIVE LOGITS
     outright
    1.10
     necessarily
    0.90
     downright
    0.85
    icably
    0.81
     technically
    0.76
     mistaken
    0.73
     extermin
    0.72
     intimidated
    0.70
     explicitly
    0.70
    ifies
    0.68
    Activations Density 0.034%

    No Known Activations